Authors: Ludi Wang, Xueqing Chen, Yi Du, Yuanchun Zhou, Yang Gao, Wenjuan Cui
Abstract: The field of catalysis holds paramount importance in shaping the trajectory of sustainable development, prompting intensive research efforts to leverage artificial intelligence (AI) in catalyst design. Presently, the fine-tuning of open-source large language models (LLMs) has yielded significant breakthroughs across various domains such as biology and healthcare. Drawing inspiration from these advancements, we introduce CataLM Cata}lytic Language Model), a large language model tailored to the domain of electrocatalytic materials. Our findings demonstrate that CataLM exhibits remarkable potential for facilitating human-AI collaboration in catalyst knowledge exploration and design. To the best of our knowledge, CataLM stands as the pioneering LLM dedicated to the catalyst domain, offering novel avenues for catalyst discovery and development.
Authors: Coenraad Mouton
Abstract: Understanding generalization in deep neural networks is an active area of research. A promising avenue of exploration has been that of margin measurements: the shortest distance to the decision boundary for a given sample or that sample's representation internal to the network. Margin-based complexity measures have been shown to be correlated with the generalization ability of deep neural networks in some circumstances but not others. The reasons behind the success or failure of these metrics are currently unclear. In this study, we examine margin-based generalization prediction methods in different settings. We motivate why these metrics sometimes fail to accurately predict generalization and how they can be improved. First, we analyze the relationship between margins measured in the input space and sample noise. We find that different types of sample noise can have a very different effect on the overall margin of a network that has modeled noisy data. Following this, we empirically evaluate how robust margins measured at different representational spaces are at predicting generalization. We find that these metrics have several limitations and that a large margin does not exhibit a strong correlation with empirical risk in many cases. Finally, we introduce a new margin-based measure that incorporates an approximation of the underlying data manifold. It is empirically demonstrated that this measure is generally more predictive of generalization than all other margin-based measures. Furthermore, we find that this measurement also outperforms other contemporary complexity measures on a well-known generalization prediction benchmark. In addition, we analyze the utility and limitations of this approach and find that this metric is well aligned with intuitions expressed in prior work.
Authors: Nienke Nijkamp, June Sallou, Niels van der Heijden, Lu\'is Cruz
Abstract: Integrating Artificial Intelligence (AI) into software systems has significantly enhanced their capabilities while escalating energy demands. Ensemble learning, combining predictions from multiple models to form a single prediction, intensifies this problem due to cumulative energy consumption. This paper presents a novel approach to model selection that addresses the challenge of balancing the accuracy of AI models with their energy consumption in a live AI ensemble system. We explore how reducing the number of models or improving the efficiency of model usage within an ensemble during inference can reduce energy demands without substantially sacrificing accuracy. This study introduces and evaluates two model selection strategies, Static and Dynamic, for optimizing ensemble learning systems performance while minimizing energy usage. Our results demonstrate that the Static strategy improves the F1 score beyond the baseline, reducing average energy usage from 100\% from the full ensemble to 6\2%. The Dynamic strategy further enhances F1 scores, using on average 76\% compared to 100% of the full ensemble. Moreover, we propose an approach that balances accuracy with resource consumption, significantly reducing energy usage without substantially impacting accuracy. This method decreased the average energy usage of the Static strategy from approximately 62\% to 14\%, and for the Dynamic strategy, from around 76\% to 57\%. Our field study of Green AI using an operational AI system developed by a large professional services provider shows the practical applicability of adopting energy-conscious model selection strategies in live production environments.
Authors: Jingchi Jiang, Rujia Shen, Boran Wang, Yi Guan
Abstract: Type 1 diabetes mellitus (T1D) is characterized by insulin deficiency and blood glucose (BG) control issues. The state-of-the-art solution for continuous BG control is reinforcement learning (RL), where an agent can dynamically adjust exogenous insulin doses in time to maintain BG levels within the target range. However, due to the lack of action guidance, the agent often needs to learn from randomized trials to understand misleading correlations between exogenous insulin doses and BG levels, which can lead to instability and unsafety. To address these challenges, we propose an introspective RL based on Counterfactual Invertible Neural Networks (CINN). We use the pre-trained CINN as a frozen introspective block of the RL agent, which integrates forward prediction and counterfactual inference to guide the policy updates, promoting more stable and safer BG control. Constructed based on interpretable causal order, CINN employs bidirectional encoders with affine coupling layers to ensure invertibility while using orthogonal weight normalization to enhance the trainability, thereby ensuring the bidirectional differentiability of network parameters. We experimentally validate the accuracy and generalization ability of the pre-trained CINN in BG prediction and counterfactual inference for action. Furthermore, our experimental results highlight the effectiveness of pre-trained CINN in guiding RL policy updates for more accurate and safer BG control.
Authors: Ziyan Yao, Fei Lin, Sheng Chai, Weijie He, Lu Dai, Xinghui Fei
Abstract: In this paper, an innovative multi-modal deep learning model is proposed to deeply integrate heterogeneous information from medical images and clinical reports. First, for medical images, convolutional neural networks were used to extract high-dimensional features and capture key visual information such as focal details, texture and spatial distribution. Secondly, for clinical report text, a two-way long and short-term memory network combined with an attention mechanism is used for deep semantic understanding, and key statements related to the disease are accurately captured. The two features interact and integrate effectively through the designed multi-modal fusion layer to realize the joint representation learning of image and text. In the empirical study, we selected a large medical image database covering a variety of diseases, combined with corresponding clinical reports for model training and validation. The proposed multimodal deep learning model demonstrated substantial superiority in the realms of disease classification, lesion localization, and clinical description generation, as evidenced by the experimental results.
Authors: Yafeng Yan, Shuyao He, Zhou Yu, Jiajie Yuan, Ziang Liu, Yan Chen
Abstract: Aiming at the limitations of traditional medical decision system in processing large-scale heterogeneous medical data and realizing highly personalized recommendation, this paper introduces a personalized medical decision algorithm utilizing graph neural network (GNN). This research innovatively integrates graph neural network technology into the medical and health field, aiming to build a high-precision representation model of patient health status by mining the complex association between patients' clinical characteristics, genetic information, living habits. In this study, medical data is preprocessed to transform it into a graph structure, where nodes represent different data entities (such as patients, diseases, genes, etc.) and edges represent interactions or relationships between entities. The core of the algorithm is to design a novel multi-scale fusion mechanism, combining the historical medical records, physiological indicators and genetic characteristics of patients, to dynamically adjust the attention allocation strategy of the graph neural network, so as to achieve highly customized analysis of individual cases. In the experimental part, this study selected several publicly available medical data sets for validation, and the results showed that compared with traditional machine learning methods and a single graph neural network model, the proposed personalized medical decision algorithm showed significantly superior performance in terms of disease prediction accuracy, treatment effect evaluation and patient risk stratification.
Authors: Chenyu Huang, Peng Ye, Tao Chen, Tong He, Xiangyu Yue, Wanli Ouyang
Abstract: The success of pretrain-finetune paradigm brings about the release of numerous model weights. In this case, merging models finetuned on different tasks to enable a single model with multi-task capabilities is gaining increasing attention for its practicability. Existing model merging methods usually suffer from (1) significant performance degradation or (2) requiring tuning by additional data or training. In this paper, we rethink and analyze the existing model merging paradigm. We discover that using a single model's weights can hardly simulate all the models' performance. To tackle this issue, we propose Elect, Mask & Rescale-Merging (EMR-Merging). We first (a) elect a unified model from all the model weights and then (b) generate extremely lightweight task-specific modulators, including masks and rescalers, to align the direction and magnitude between the unified model and each specific model, respectively. EMR-Merging is tuning-free, thus requiring no data availability or any additional training while showing impressive performance. We find that EMR-Merging shows outstanding performance compared to existing merging methods under different classical and newly-established settings, including merging different numbers of vision models (up to 30), NLP models, PEFT models, and multi-modal models.
Authors: Hanlin Gu, WinKent Ong, Chee Seng Chan, Lixin Fan
Abstract: The advent of Federated Learning (FL) highlights the practical necessity for the 'right to be forgotten' for all clients, allowing them to request data deletion from the machine learning model's service provider. This necessity has spurred a growing demand for Federated Unlearning (FU). Feature unlearning has gained considerable attention due to its applications in unlearning sensitive features, backdoor features, and bias features. Existing methods employ the influence function to achieve feature unlearning, which is impractical for FL as it necessitates the participation of other clients in the unlearning process. Furthermore, current research lacks an evaluation of the effectiveness of feature unlearning. To address these limitations, we define feature sensitivity in the evaluation of feature unlearning according to Lipschitz continuity. This metric characterizes the rate of change or sensitivity of the model output to perturbations in the input feature. We then propose an effective federated feature unlearning framework called Ferrari, which minimizes feature sensitivity. Extensive experimental results and theoretical analysis demonstrate the effectiveness of Ferrari across various feature unlearning scenarios, including sensitive, backdoor, and biased features.
Authors: Xiaoling Zhou, Ou Wu, Michael K. Ng, Hao Jiang
Abstract: Data valuation has garnered increasing attention in recent years, given the critical role of high-quality data in various applications, particularly in machine learning tasks. There are diverse technical avenues to quantify the value of data within a corpus. While Shapley value-based methods are among the most widely used techniques in the literature due to their solid theoretical foundation, the accurate calculation of Shapley values is often intractable, leading to the proposal of numerous approximated calculation methods. Despite significant progress, nearly all existing methods overlook the utilization of distribution information of values within a data corpus. In this paper, we demonstrate that both global and local statistical information of value distributions hold significant potential for data valuation within the context of machine learning. Firstly, we explore the characteristics of both global and local value distributions across several simulated and real data corpora. Useful observations and clues are obtained. Secondly, we propose a new data valuation method that estimates Shapley values by incorporating the explored distribution characteristics into an existing method, AME. Thirdly, we present a new path to address the dynamic data valuation problem by formulating an optimization problem that integrates information of both global and local value distributions. Extensive experiments are conducted on Shapley value estimation, value-based data removal/adding, mislabeled data detection, and incremental/decremental data valuation. The results showcase the effectiveness and efficiency of our proposed methodologies, affirming the significant potential of global and local value distributions in data valuation.
Authors: Aashu, Kanchan Rajwar, Millie Pant, Kusum Deep
Abstract: Food production is a vital global concern and the potential for an agritech revolution through artificial intelligence (AI) remains largely unexplored. This paper presents a comprehensive review focused on the application of machine learning (ML) in agriculture, aiming to explore its transformative potential in farming practices and efficiency enhancement. To understand the extent of research activity in this field, statistical data have been gathered, revealing a substantial growth trend in recent years. This indicates that it stands out as one of the most dynamic and vibrant research domains. By introducing the concept of ML and delving into the realm of smart agriculture, including Precision Agriculture, Smart Farming, Digital Agriculture, and Agriculture 4.0, we investigate how AI can optimize crop output and minimize environmental impact. We highlight the capacity of ML to analyze and classify agricultural data, providing examples of improved productivity and profitability on farms. Furthermore, we discuss prominent ML models and their unique features that have shown promising results in agricultural applications. Through a systematic review of the literature, this paper addresses the existing literature gap on AI in agriculture and offers valuable information to newcomers and researchers. By shedding light on unexplored areas within this emerging field, our objective is to facilitate a deeper understanding of the significant contributions and potential of AI in agriculture, ultimately benefiting the research community.
Authors: Long Le, Marcel Hussing, Eric Eaton
Abstract: This work studies the intersection of continual and federated learning, in which independent agents face unique tasks in their environments and incrementally develop and share knowledge. We introduce a mathematical framework capturing the essential aspects of distributed continual learning, including agent model and statistical heterogeneity, continual distribution shift, network topology, and communication constraints. Operating on the thesis that distributed continual learning enhances individual agent performance over single-agent learning, we identify three modes of information exchange: data instances, full model parameters, and modular (partial) model parameters. We develop algorithms for each sharing mode and conduct extensive empirical investigations across various datasets, topology structures, and communication limits. Our findings reveal three key insights: sharing parameters is more efficient than sharing data as tasks become more complex; modular parameter sharing yields the best performance while minimizing communication costs; and combining sharing modes can cumulatively improve performance.
Authors: Juan Soto, Ram\'on Carmenaty, Miguel Lastra, Juan M. Fern\'andez-Luna, Jos\'e M. Ben\'itez
Abstract: Customer segmentation is a fundamental process to develop effective marketing strategies, personalize customer experience and boost their retention and loyalty. This problem has been widely addressed in the scientific literature, yet no definitive solution for every case is available. A specific case study characterized by several individualizing features is thoroughly analyzed and discussed in this paper. Because of the case properties a robust and innovative approach to both data handling and analytical processes is required. The study led to a sound proposal for customer segmentation. The highlights of the proposal include a convenient data partition to decompose the problem, an adaptive distance function definition and its optimization through genetic algorithms. These comprehensive data handling strategies not only enhance the dataset reliability for segmentation analysis but also support the operational efficiency and marketing strategies of sports centers, ultimately improving the customer experience.
Authors: Xishun Liao, Brian Yueshuai He, Qinhua Jiang, Chenchen Kuai, Jiaqi Ma
Abstract: Human mobility significantly impacts various aspects of society, including transportation, urban planning, and public health. The increasing availability of diverse mobility data and advancements in deep learning have revolutionized mobility modeling. Existing deep learning models, however, mainly study spatio-temporal patterns using trajectories and often fall short in capturing the underlying semantic interdependency among activities. Moreover, they are also constrained by the data source. These two factors thereby limit their realism and adaptability, respectively. Meanwhile, traditional activity-based models (ABMs) in transportation modeling rely on rigid assumptions and are costly and time-consuming to calibrate, making them difficult to adapt and scale to new regions, especially those regions with limited amount of required conventional travel data. To address these limitations, we develop a novel generative deep learning approach for human mobility modeling and synthesis, using ubiquitous and open-source data. Additionally, the model can be fine-tuned with local data, enabling adaptable and accurate representations of mobility patterns across different regions. The model is evaluated on a nationwide dataset of the United States, where it demonstrates superior performance in generating activity chains that closely follow ground truth distributions. Further tests using state- or city-specific datasets from California, Washington, and Mexico City confirm its transferability. This innovative approach offers substantial potential to advance mobility modeling research, especially in generating human activity chains as input for downstream activity-based mobility simulation models and providing enhanced tools for urban planners and policymakers.
Authors: Pranjol Sen Gupta, Md Rajib Hossen, Pengfei Li, Shaolei Ren, Mohammad A. Islam
Abstract: Freshwater scarcity is a global problem that requires collective efforts across all industry sectors. Nevertheless, a lack of access to operational water footprint data bars many applications from exploring optimization opportunities hidden within the temporal and spatial variations. To break this barrier into research in water sustainability, we build a dataset for operation direct water usage in the cooling systems and indirect water embedded in electricity generation. Our dataset consists of the hourly water efficiency of major U.S. cities and states from 2019 to 2023. We also offer cooling system models that capture the impact of weather on water efficiency. We present a preliminary analysis of our dataset and discuss three potential applications that can benefit from it. Our dataset is publicly available at Open Science Framework (OSF)
Authors: Yanshu Wang, Wenyang He, Tong Yang
Abstract: Large Language Models (LLMs) have significantly advanced natural language processing tasks such as machine translation, text generation, and sentiment analysis. However, their large size, often consisting of billions of parameters, poses challenges for storage, computation, and deployment, particularly in resource-constrained environments like mobile devices and edge computing platforms. Effective compression and quantization techniques are crucial for addressing these issues, reducing memory footprint and computational requirements without significantly compromising performance. Traditional methods that uniformly map parameters to compressed spaces fail to account for the uneven distribution of parameters, leading to substantial accuracy loss. In this work, we propose Athena, a novel algorithm for efficient block-wise post-training quantization of LLMs. Athena leverages Second-Order Matrix Derivative Information to guide the quantization process using the curvature information of the loss landscape. By grouping parameters by columns or rows and iteratively optimizing the quantization process, Athena updates the model parameters and Hessian matrix to achieve significant compression while maintaining high accuracy. This makes Athena a practical solution for deploying LLMs in various settings.
Authors: Sheng Yue, Xingyuan Hua, Lili Chen, Ju Ren
Abstract: Federated Reinforcement Learning (FRL) has garnered increasing attention recently. However, due to the intrinsic spatio-temporal non-stationarity of data distributions, the current approaches typically suffer from high interaction and communication costs. In this paper, we introduce a new FRL algorithm, named \alg{}, that utilizes momentum, importance sampling, and additional server-side adjustment to control the shift of stochastic policy gradients and enhance the efficiency of data utilization. We prove that by proper selection of momentum parameters and interaction frequency, \alg{} can achieve $\tilde{\mathcal{O}}(H N^{-1}\epsilon^{-3/2})$ and $\tilde{\mathcal{O}}(\epsilon^{-1})$ interaction and communication complexities ($N$ represents the number of agents), where the interaction complexity achieves linear speedup with the number of agents, and the communication complexity aligns the best achievable of existing first-order FL algorithms. Extensive experiments corroborate the substantial performance gains of \alg{} over existing methods on a suite of complex and high-dimensional benchmarks.
Authors: Kai Huang, Wei Gao
Abstract: Text-to-image diffusion models can be fine-tuned in custom domains to adapt to specific user preferences, but such unconstrained adaptability has also been utilized for illegal purposes, such as forging public figures' portraits and duplicating copyrighted artworks. Most existing work focuses on detecting the illegally generated contents, but cannot prevent or mitigate illegal adaptations of diffusion models. Other schemes of model unlearning and reinitialization, similarly, cannot prevent users from relearning the knowledge of illegal model adaptation with custom data. In this paper, we present FreezeAsGuard, a new technique that addresses these limitations and enables irreversible mitigation of illegal adaptations of diffusion models. The basic approach is that the model publisher selectively freezes tensors in pre-trained diffusion models that are critical to illegal model adaptations, to mitigate the fine-tuned model's representation power in illegal domains but minimize the impact on legal model adaptations in other domains. Such tensor freezing can be enforced via APIs provided by the model publisher for fine-tuning, can motivate users' adoption due to its computational savings. Experiment results with datasets in multiple domains show that FreezeAsGuard provides stronger power in mitigating illegal model adaptations of generating fake public figures' portraits, while having the minimum impact on model adaptation in other legal domains. The source code is available at: https://github.com/pittisl/FreezeAsGuard/
Authors: Tao Zou, Yuhao Mao, Junchen Ye, Bowen Du
Abstract: Dynamic graph learning equips the edges with time attributes and allows multiple links between two nodes, which is a crucial technology for understanding evolving data scenarios like traffic prediction and recommendation systems. Existing works obtain the evolving patterns mainly depending on the most recent neighbor sequences. However, we argue that whether two nodes will have interaction with each other in the future is highly correlated with the same interaction that happened in the past. Only considering the recent neighbors overlooks the phenomenon of repeat behavior and fails to accurately capture the temporal evolution of interactions. To fill this gap, this paper presents RepeatMixer, which considers evolving patterns of first and high-order repeat behavior in the neighbor sampling strategy and temporal information learning. Firstly, we define the first-order repeat-aware nodes of the source node as the destination nodes that have interacted historically and extend this concept to high orders as nodes in the destination node's high-order neighbors. Then, we extract neighbors of the source node that interacted before the appearance of repeat-aware nodes with a slide window strategy as its neighbor sequence. Next, we leverage both the first and high-order neighbor sequences of source and destination nodes to learn temporal patterns of interactions via an MLP-based encoder. Furthermore, considering the varying temporal patterns on different orders, we introduce a time-aware aggregation mechanism that adaptively aggregates the temporal representations from different orders based on the significance of their interaction time sequences. Experimental results demonstrate the superiority of RepeatMixer over state-of-the-art models in link prediction tasks, underscoring the effectiveness of the proposed repeat-aware neighbor sampling strategy.
Authors: Sheng Yue, Zerui Qin, Xingyuan Hua, Yongheng Deng, Ju Ren
Abstract: Federated Reinforcement Learning (FRL) has been deemed as a promising solution for intelligent decision-making in the era of Artificial Internet of Things. However, existing FRL approaches often entail repeated interactions with the environment during local updating, which can be prohibitively expensive or even infeasible in many real-world domains. To overcome this challenge, this paper proposes a novel offline federated policy optimization algorithm, named $\texttt{DRPO}$, which enables distributed agents to collaboratively learn a decision policy only from private and static data without further environmental interactions. $\texttt{DRPO}$ leverages dual regularization, incorporating both the local behavioral policy and the global aggregated policy, to judiciously cope with the intrinsic two-tier distributional shifts in offline FRL. Theoretical analysis characterizes the impact of the dual regularization on performance, demonstrating that by achieving the right balance thereof, $\texttt{DRPO}$ can effectively counteract distributional shifts and ensure strict policy improvement in each federative learning round. Extensive experiments validate the significant performance gains of $\texttt{DRPO}$ over baseline methods.
Authors: Sheng Yue, Jiani Liu, Xingyuan Hua, Ju Ren, Sen Lin, Junshan Zhang, Yaoxue Zhang
Abstract: Offline Imitation Learning (IL) with imperfect demonstrations has garnered increasing attention owing to the scarcity of expert data in many real-world domains. A fundamental problem in this scenario is how to extract positive behaviors from noisy data. In general, current approaches to the problem select data building on state-action similarity to given expert demonstrations, neglecting precious information in (potentially abundant) $\textit{diverse}$ state-actions that deviate from expert ones. In this paper, we introduce a simple yet effective data selection method that identifies positive behaviors based on their resultant states -- a more informative criterion enabling explicit utilization of dynamics information and effective extraction of both expert and beneficial diverse behaviors. Further, we devise a lightweight behavior cloning algorithm capable of leveraging the expert and selected data correctly. In the experiments, we evaluate our method on a suite of complex and high-dimensional offline IL benchmarks, including continuous-control and vision-based tasks. The results demonstrate that our method achieves state-of-the-art performance, outperforming existing methods on $\textbf{20/21}$ benchmarks, typically by $\textbf{2-5x}$, while maintaining a comparable runtime to Behavior Cloning ($\texttt{BC}$).
Authors: Sheng Yue, Xingyuan Hua, Ju Ren, Sen Lin, Junshan Zhang, Yaoxue Zhang
Abstract: In this paper, we study offline-to-online Imitation Learning (IL) that pretrains an imitation policy from static demonstration data, followed by fast finetuning with minimal environmental interaction. We find the na\"ive combination of existing offline IL and online IL methods tends to behave poorly in this context, because the initial discriminator (often used in online IL) operates randomly and discordantly against the policy initialization, leading to misguided policy optimization and $\textit{unlearning}$ of pretraining knowledge. To overcome this challenge, we propose a principled offline-to-online IL method, named $\texttt{OLLIE}$, that simultaneously learns a near-expert policy initialization along with an $\textit{aligned discriminator initialization}$, which can be seamlessly integrated into online IL, achieving smooth and fast finetuning. Empirically, $\texttt{OLLIE}$ consistently and significantly outperforms the baseline methods in $\textbf{20}$ challenging tasks, from continuous control to vision-based domains, in terms of performance, demonstration efficiency, and convergence speed. This work may serve as a foundation for further exploration of pretraining and finetuning in the context of IL.
Authors: Yihang Wang, Yuying Qiu, Peng Chen, Kai Zhao, Yang Shu, Zhongwen Rao, Lujia Pan, Bin Yang, Chenjuan Guo
Abstract: With the increasing collection of time series data from various domains, there arises a strong demand for general time series forecasting models pre-trained on a large number of time-series datasets to support a variety of downstream prediction tasks. Enabling general time series forecasting faces two challenges: how to obtain unified representations from multi-domian time series data, and how to capture domain-specific features from time series data across various domains for adaptive transfer in downstream tasks. To address these challenges, we propose a Register Assisted General Time Series Forecasting Model with Decomposed Frequency Learning (ROSE), a novel pre-trained model for time series forecasting. ROSE employs Decomposed Frequency Learning for the pre-training task, which decomposes coupled semantic and periodic information in time series with frequency-based masking and reconstruction to obtain unified representations across domains. We also equip ROSE with a Time Series Register, which learns to generate a register codebook to capture domain-specific representations during pre-training and enhances domain-adaptive transfer by selecting related register tokens on downstream tasks. After pre-training on large-scale time series data, ROSE achieves state-of-the-art forecasting performance on 8 real-world benchmarks. Remarkably, even in few-shot scenarios, it demonstrates competitive or superior performance compared to existing methods trained with full data.
Authors: Zhangchen Zhou, Yaoyu Zhang, Zhi-Qin John Xu
Abstract: Grokking is the phenomenon where neural networks NNs initially fit the training data and later generalize to the test data during training. In this paper, we empirically provide a frequency perspective to explain the emergence of this phenomenon in NNs. The core insight is that the networks initially learn the less salient frequency components present in the test data. We observe this phenomenon across both synthetic and real datasets, offering a novel viewpoint for elucidating the grokking phenomenon by characterizing it through the lens of frequency dynamics during the training process. Our empirical frequency-based analysis sheds new light on understanding the grokking phenomenon and its underlying mechanisms.
Authors: Deepak Narayan Gadde, Sebastian Simon, Djones Lettnin, Thomas Ziller
Abstract: The verification throughput is becoming a major challenge bottleneck, since the complexity and size of SoC designs are still ever increasing. Simply adding more CPU cores and running more tests in parallel will not scale anymore. This paper discusses various methods of improving verification throughput: ranking and the new machine learning (ML) based technology introduced by Cadence i.e. Xcelium ML. Both methods aim at getting comparable coverage in less CPU time by applying more efficient stimulus. Ranking selects specific seeds that simply turned out to come up with the largest coverage in previous simulations, while Xcelium ML generates optimized patterns as a result of finding correlations between randomization points and achieved coverage of previous regressions. Quantified results as well as pros & cons of each approach are discussed in this paper at the example of three actual industry projects. Both Xcelium ML and Ranking methods gave comparable compression & speedup factors around 3 consistently. But the optimized ML based regressions simulated new random scenarios occasionally producing a coverage regain of more than 100%. Finally, a methodology is proposed to use Xcelium ML efficiently throughout the product development.
Authors: Shen Yuan, Haotian Liu, Hongteng Xu
Abstract: While following different technical routes, both low-rank and orthogonal adaptation techniques can efficiently adapt large-scale pre-training models in specific tasks or domains based on a small piece of trainable parameters. In this study, we bridge the gap between these two techniques, proposing a simple but effective adaptation method based on Householder reflections. Given a pre-trained model, our method fine-tunes its layers by multiplying each frozen weight matrix with an orthogonal matrix constructed by a chain of learnable Householder reflections (HRs). This HR-based orthogonal fine-tuning is equivalent to an adaptive low-rank adaptation. Moreover, we show that the orthogonality of the reflection planes corresponding to the HRs impacts the model capacity and regularity. The analysis motivates us to regularize the orthogonality of the HRs, leading to different implementations of the proposed Householder reflection adaptation (HRA) method. Compared with state-of-the-art methods, HRA achieves superior performance with fewer learnable parameters when adapting large language models and conditional image generators. The code is available at https://github.com/DaShenZi721/HRA
Authors: Xiangrui Xu, Qiao Zhang, Rui Ning, Chunsheng Xin, Hongyi Wu
Abstract: The prevalent use of Transformer-like models, exemplified by ChatGPT in modern language processing applications, underscores the critical need for enabling private inference essential for many cloud-based services reliant on such models. However, current privacy-preserving frameworks impose significant communication burden, especially for non-linear computation in Transformer model. In this paper, we introduce a novel plug-in method Comet to effectively reduce the communication cost without compromising the inference performance. We second introduce an efficient approximation method to eliminate the heavy communication in finding good initial approximation. We evaluate our Comet on Bert and RoBERTa models with GLUE benchmark datasets, showing up to 3.9$\times$ less communication and 3.5$\times$ speedups while keep competitive model performance compared to the prior art.
Authors: Elad Liebman
Abstract: Fault detection is a key challenge in the management of complex systems. In the context of SparkCognition's efforts towards predictive maintenance in large scale industrial systems, this problem is often framed in terms of anomaly detection - identifying patterns of behavior in the data which deviate from normal. Patterns of normal behavior aren't captured simply in the coarse statistics of measured signals. Rather, the multivariate sequential pattern itself can be indicative of normal vs. abnormal behavior. For this reason, normal behavior modeling that relies on snapshots of the data without taking into account temporal relationships as they evolve would be lacking. However, common strategies for dealing with temporal dependence, such as Recurrent Neural Networks or attention mechanisms are oftentimes computationally expensive and difficult to train. In this paper, we propose a fast and efficient approach to anomaly detection and alert filtering based on sequential pattern similarities. In our empirical analysis section, we show how this approach can be leveraged for a variety of purposes involving anomaly detection on a large scale real-world industrial system. Subsequently, we test our approach on a publicly-available dataset in order to establish its general applicability and robustness compared to a state-of-the-art baseline. We also demonstrate an efficient way of optimizing the framework based on an alert recall objective function.
Authors: Ziao Yang, Han Yue, Jian Chen, Hongfu Liu
Abstract: Shapley value-based data valuation methods, originating from cooperative game theory, quantify the usefulness of each individual sample by considering its contribution to all possible training subsets. Despite their extensive applications, these methods encounter the challenge of value inflation - while samples with negative Shapley values are detrimental, some with positive values can also be harmful. This challenge prompts two fundamental questions: the suitability of zero as a threshold for distinguishing detrimental from beneficial samples and the determination of an appropriate threshold. To address these questions, we focus on KNN-Shapley and propose Calibrated KNN-Shapley (CKNN-Shapley), which calibrates zero as the threshold to distinguish detrimental samples from beneficial ones by mitigating the negative effects of small-sized training subsets. Through extensive experiments, we demonstrate the effectiveness of CKNN-Shapley in alleviating data valuation inflation, detecting detrimental samples, and assessing data quality. We also extend our approach beyond conventional classification settings, applying it to diverse and practical scenarios such as learning with mislabeled data, online learning with stream data, and active learning for label annotation.
Authors: Ziao Yang, Han Yue, Jian Chen, Hongfu Liu
Abstract: Influence functions serve as crucial tools for assessing sample influence in model interpretation, subset training set selection, noisy label detection, and more. By employing the first-order Taylor extension, influence functions can estimate sample influence without the need for expensive model retraining. However, applying influence functions directly to deep models presents challenges, primarily due to the non-convex nature of the loss function and the large size of model parameters. This difficulty not only makes computing the inverse of the Hessian matrix costly but also renders it non-existent in some cases. Various approaches, including matrix decomposition, have been explored to expedite and approximate the inversion of the Hessian matrix, with the aim of making influence functions applicable to deep models. In this paper, we revisit a specific, albeit naive, yet effective approximation method known as TracIn. This method substitutes the inverse of the Hessian matrix with an identity matrix. We provide deeper insights into why this simple approximation method performs well. Furthermore, we extend its applications beyond measuring model utility to include considerations of fairness and robustness. Finally, we enhance TracIn through an ensemble strategy. To validate its effectiveness, we conduct experiments on synthetic data and extensive evaluations on noisy label detection, sample selection for large language model fine-tuning, and defense against adversarial attacks.
Authors: Ziyan Wang, Mohamed Ragab, Wenmian Yang, Min Wu, Sinno Jialin Pan, Jie Zhang, Zhenghua Chen
Abstract: Unsupervised domain adaptation (UDA) has achieved remarkable success in fault diagnosis, bringing significant benefits to diverse industrial applications. While most UDA methods focus on cross-working condition scenarios where the source and target domains are notably similar, real-world applications often grapple with severe domain shifts. We coin the term `distant domain adaptation problem' to describe the challenge of adapting from a labeled source domain to a significantly disparate unlabeled target domain. This problem exhibits the risk of negative transfer, where extraneous knowledge from the source domain adversely affects the target domain performance. Unfortunately, conventional UDA methods often falter in mitigating this negative transfer, leading to suboptimal performance. In response to this challenge, we propose a novel Online Selective Adversarial Alignment (OSAA) approach. Central to OSAA is its ability to dynamically identify and exclude distant source samples via an online gradient masking approach, focusing primarily on source samples that closely resemble the target samples. Furthermore, recognizing the inherent complexities in bridging the source and target domains, we construct an intermediate domain to act as a transitional domain and ease the adaptation process. Lastly, we develop a class-conditional adversarial adaptation to address the label distribution disparities while learning domain invariant representation to account for potential label distribution disparities between the domains. Through detailed experiments and ablation studies on two real-world datasets, we validate the superior performance of the OSAA method over state-of-the-art methods, underscoring its significant utility in practical scenarios with severe domain shifts.
Authors: Ryan Benkert, Mohit Prabhushankar, Ghassan AlRegib
Abstract: In this paper, we discuss feature engineering for single-pass uncertainty estimation. For accurate uncertainty estimates, neural networks must extract differences in the feature space that quantify uncertainty. This could be achieved by current single-pass approaches that maintain feature distances between data points as they traverse the network. While initial results are promising, maintaining feature distances within the network representations frequently inhibits information compression and opposes the learning objective. We study this effect theoretically and empirically to arrive at a simple conclusion: preserving feature distances in the output is beneficial when the preserved features contribute to learning the label distribution and act in opposition otherwise. We then propose Transitional Uncertainty with Layered Intermediate Predictions (TULIP) as a simple approach to address the shortcomings of current single-pass estimators. Specifically, we implement feature preservation by extracting features from intermediate representations before information is collapsed by subsequent layers. We refer to the underlying preservation mechanism as transitional feature preservation. We show that TULIP matches or outperforms current single-pass methods on standard benchmarks and in practical settings where these methods are less reliable (imbalances, complex architectures, medical modalities).
Authors: Mang Ye, Wei Shen, Eduard Snezhko, Vassili Kovalev, Pong C. Yuen, Bo Du
Abstract: Vertical Federated Learning (VFL) is a privacy-preserving distributed learning paradigm where different parties collaboratively learn models using partitioned features of shared samples, without leaking private data. Recent research has shown promising results addressing various challenges in VFL, highlighting its potential for practical applications in cross-domain collaboration. However, the corresponding research is scattered and lacks organization. To advance VFL research, this survey offers a systematic overview of recent developments. First, we provide a history and background introduction, along with a summary of the general training protocol of VFL. We then revisit the taxonomy in recent reviews and analyze limitations in-depth. For a comprehensive and structured discussion, we synthesize recent research from three fundamental perspectives: effectiveness, security, and applicability. Finally, we discuss several critical future research directions in VFL, which will facilitate the developments in this field. We provide a collection of research lists and periodically update them at https://github.com/shentt67/VFL_Survey.
Authors: M. Saeid HaghighiFard, Sinem Coleri
Abstract: Hierarchical Federated Learning (HFL) faces the significant challenge of adversarial or unreliable vehicles in vehicular networks, which can compromise the model's integrity through misleading updates. Addressing this, our study introduces a novel framework that integrates dynamic vehicle selection and robust anomaly detection mechanisms, aiming to optimize participant selection and mitigate risks associated with malicious contributions. Our approach involves a comprehensive vehicle reliability assessment, considering historical accuracy, contribution frequency, and anomaly records. An anomaly detection algorithm is utilized to identify anomalous behavior by analyzing the cosine similarity of local or model parameters during the federated learning (FL) process. These anomaly records are then registered and combined with past performance for accuracy and contribution frequency to identify the most suitable vehicles for each learning round. Dynamic client selection and anomaly detection algorithms are deployed at different levels, including cluster heads (CHs), cluster members (CMs), and the Evolving Packet Core (EPC), to detect and filter out spurious updates. Through simulation-based performance evaluation, our proposed algorithm demonstrates remarkable resilience even under intense attack conditions. Even in the worst-case scenarios, it achieves convergence times at $63$\% as effective as those in scenarios without any attacks. Conversely, in scenarios without utilizing our proposed algorithm, there is a high likelihood of non-convergence in the FL process.
Authors: Leyang Zhang, Yaoyu Zhang, Tao Luo
Abstract: This paper presents a comprehensive analysis of critical point sets in two-layer neural networks. To study such complex entities, we introduce the critical embedding operator and critical reduction operator as our tools. Given a critical point, we use these operators to uncover the whole underlying critical set representing the same output function, which exhibits a hierarchical structure. Furthermore, we prove existence of saddle branches for any critical set whose output function can be represented by a narrower network. Our results provide a solid foundation to the further study of optimization and training behavior of neural networks.
Authors: Ziming Liu, Longjian Liu, Robert E. Heidel, Xiaopeng Zhao
Abstract: This article uses machine learning (ML) and explainable artificial intelligence (XAI) techniques to investigate the relationship between nutritional status and mortality rates associated with Alzheimers disease (AD). The Third National Health and Nutrition Examination Survey (NHANES III) database is employed for analysis. The random forest model is selected as the base model for XAI analysis, and the Shapley Additive Explanations (SHAP) method is used to assess feature importance. The results highlight significant nutritional factors such as serum vitamin B12 and glycated hemoglobin. The study demonstrates the effectiveness of random forests in predicting AD mortality compared to other diseases. This research provides insights into the impact of nutrition on AD and contributes to a deeper understanding of disease progression.
Authors: Tingting Chen, Shijing Si
Abstract: Housing has emerged as a crucial concern among young individuals residing in major cities, including Shanghai. Given the unprecedented surge in property prices in this metropolis, young people have increasingly resorted to the rental market to address their housing needs. This study utilizes five traditional machine learning methods: multiple linear regression (MLR), ridge regression (RR), lasso regression (LR), decision tree (DT), and random forest (RF), along with a Large Language Model (LLM) approach using ChatGPT, for predicting the rental prices of lane houses in Shanghai. It applies these methods to examine a public data sample of about 2,609 lane house rental transactions in 2021 in Shanghai, and then compares the results of these methods. In terms of predictive power, RF has achieved the best performance among the traditional methods. However, the LLM approach, particularly in the 10-shot scenario, shows promising results that surpass traditional methods in terms of R-Squared value. The three performance metrics: mean squared error (MSE), mean absolute error (MAE), and R-Squared, are used to evaluate the models. Our conclusion is that while traditional machine learning models offer robust techniques for rental price prediction, the integration of LLM such as ChatGPT holds significant potential for enhancing predictive accuracy.
Authors: Joshua Offergeld, Marcel van Gerven, Nasir Ahmad
Abstract: A significant increase in the commercial use of deep neural network models increases the need for efficient AI. Node pruning is the art of removing computational units such as neurons, filters, attention heads, or even entire layers while keeping network performance at a maximum. This can significantly reduce the inference time of a deep network and thus enhance its efficiency. Few of the previous works have exploited the ability to recover performance by reorganizing network parameters while pruning. In this work, we propose to create a subspace from unit activations which enables node pruning while recovering maximum accuracy. We identify that for effective node pruning, a subspace can be created using a triangular transformation matrix, which we show to be equivalent to Gram-Schmidt orthogonalization, which automates this procedure. We further improve this method by reorganizing the network prior to subspace formation. Finally, we leverage the orthogonal subspaces to identify layer-wise pruning ratios appropriate to retain a significant amount of the layer-wise information. We show that this measure outperforms existing pruning methods on VGG networks. We further show that our method can be extended to other network architectures such as residual networks.
Authors: ChungYi Lin, Shen-Lung Tung, Hung-Ting Su, Winston H. Hsu
Abstract: Traditional traffic prediction, limited by the scope of sensor data, falls short in comprehensive traffic management. Mobile networks offer a promising alternative using network activity counts, but these lack crucial directionality. Thus, we present the TeltoMob dataset, featuring undirected telecom counts and corresponding directional flows, to predict directional mobility flows on roadways. To address this, we propose a two-stage spatio-temporal graph neural network (STGNN) framework. The first stage uses a pre-trained STGNN to process telecom data, while the second stage integrates directional and geographic insights for accurate prediction. Our experiments demonstrate the framework's compatibility with various STGNN models and confirm its effectiveness. We also show how to incorporate the framework into real-world transportation systems, enhancing sustainable urban mobility.
Authors: Linglong Qian, Zina Ibrahim, Wenjie Du, Yiyuan Yang, Richard JB Dobson
Abstract: In this study, we explore the impact of different masking strategies on time series imputation models. We evaluate the effects of pre-masking versus in-mini-batch masking, normalization timing, and the choice between augmenting and overlaying artificial missingness. Using three diverse datasets, we benchmark eleven imputation models with different missing rates. Our results demonstrate that masking strategies significantly influence imputation accuracy, revealing that more sophisticated and data-driven masking designs are essential for robust model evaluation. We advocate for refined experimental designs and comprehensive disclosureto better simulate real-world patterns, enhancing the practical applicability of imputation models.
Authors: Ze Cheng, Zhongkai Hao, Xiaoqiang Wang, Jianing Huang, Youjia Wu, Xudan Liu, Yiru Zhao, Songming Liu, Hang Su
Abstract: For partial differential equations on domains of arbitrary shapes, existing works of neural operators attempt to learn a mapping from geometries to solutions. It often requires a large dataset of geometry-solution pairs in order to obtain a sufficiently accurate neural operator. However, for many industrial applications, e.g., engineering design optimization, it can be prohibitive to satisfy the requirement since even a single simulation may take hours or days of computation. To address this issue, we propose reference neural operators (RNO), a novel way of implementing neural operators, i.e., to learn the smooth dependence of solutions on geometric deformations. Specifically, given a reference solution, RNO can predict solutions corresponding to arbitrary deformations of the referred geometry. This approach turns out to be much more data efficient. Through extensive experiments, we show that RNO can learn the dependence across various types and different numbers of geometry objects with relatively small datasets. RNO outperforms baseline models in accuracy by a large lead and achieves up to 80% error reduction.
Authors: Zhoujie Ding, Ken Ziyu Liu, Pura Peetathawatchai, Berivan Isik, Sanmi Koyejo
Abstract: Low-rank adaptation of large models, particularly LoRA, has gained traction due to its computational efficiency. This efficiency, contrasted with the prohibitive costs of full-model fine-tuning, means that practitioners often turn to LoRA and sometimes without a complete understanding of its ramifications. In this study, we focus on fairness and ask whether LoRA has an unexamined impact on utility, calibration, and resistance to membership inference across different subgroups (e.g., genders, races, religions) compared to a full-model fine-tuning baseline. We present extensive experiments across vision and language domains and across classification and generation tasks using ViT-Base, Swin-v2-Large, Llama-2 7B, and Mistral 7B. Intriguingly, experiments suggest that while one can isolate cases where LoRA exacerbates model bias across subgroups, the pattern is inconsistent -- in many cases, LoRA has equivalent or even improved fairness compared to the base model or its full fine-tuning baseline. We also examine the complications of evaluating fine-tuning fairness relating to task design and model token bias, calling for more careful fairness evaluations in future work.
Authors: Louis Fournier (MLIA), Adel Nabli (MLIA, Mila), Masih Aminbeidokhti (ETS), Marco Pedersoli (ETS), Eugene Belilovsky (Mila), Edouard Oyallon
Abstract: The performance of deep neural networks is enhanced by ensemble methods, which average the output of several models. However, this comes at an increased cost at inference. Weight averaging methods aim at balancing the generalization of ensembling and the inference speed of a single model by averaging the parameters of an ensemble of models. Yet, naive averaging results in poor performance as models converge to different loss basins, and aligning the models to improve the performance of the average is challenging. Alternatively, inspired by distributed training, methods like DART and PAPA have been proposed to train several models in parallel such that they will end up in the same basin, resulting in good averaging accuracy. However, these methods either compromise ensembling accuracy or demand significant communication between models during training. In this paper, we introduce WASH, a novel distributed method for training model ensembles for weight averaging that achieves state-of-the-art image classification accuracy. WASH maintains models within the same basin by randomly shuffling a small percentage of weights during training, resulting in diverse models and lower communication costs compared to standard parameter averaging methods.
Authors: Xi Zhu, Songcan Yu, Junbo Wang, Qinglin Yang
Abstract: Federated learning (FL), as an emerging collaborative learning paradigm, has garnered significant attention due to its capacity to preserve privacy within distributed learning systems. In these systems, clients collaboratively train a unified neural network model using their local datasets and share model parameters rather than raw data, enhancing privacy. Predominantly, FL systems are designed for mobile and edge computing environments where training typically occurs over wireless networks. Consequently, as model sizes increase, the conventional FL frameworks increasingly consume substantial communication resources. To address this challenge and improve communication efficiency, this paper introduces a novel hierarchical FL framework that integrates the benefits of clustered FL and model compression. We present an adaptive clustering algorithm that identifies a core client and dynamically organizes clients into clusters. Furthermore, to enhance transmission efficiency, each core client implements a local aggregation with compression (LC aggregation) algorithm after collecting compressed models from other clients within the same cluster. Simulation results affirm that our proposed algorithms not only maintain comparable predictive accuracy but also significantly reduce energy consumption relative to existing FL mechanisms.
Authors: Xiangyu Dong, Xingyi Zhang, Yanni Sun, Lei Chen, Mingxuan Yuan, Sibo Wang
Abstract: The smoothing issue leads to indistinguishable node representations, which poses a significant challenge in the field of graph learning. However, this issue also presents an opportunity to reveal underlying properties behind different types of nodes, which have been overlooked in previous studies. Through empirical and theoretical analysis of real-world node anomaly detection (NAD) datasets, we observe that anomalous and normal nodes show different patterns in the smoothing process, which can be leveraged to enhance NAD tasks. Motivated by these findings, in this paper, we propose a novel unsupervised NAD framework. Specifically, according to our theoretical analysis, we design a Smoothing Learning Component. Subsequently, we introduce a Smoothing-aware Spectral Graph Neural Network, which establishes the connection between the spectral space of graphs and the smoothing process. Additionally, we demonstrate that the Dirichlet Energy, which reflects the smoothness of a graph, can serve as coefficients for node representations across different dimensions of the spectral space. Building upon these observations and analyses, we devise a novel anomaly measure for the NAD task. Extensive experiments on 9 real-world datasets show that SmoothGNN outperforms the best rival by an average of 14.66% in AUC and 7.28% in Precision, with 75x running time speed-up, which validates the effectiveness and efficiency of our framework.
Authors: Zhou Hang, Yuezhou Ma, Haixu Wu, Haowen Wang, Mingsheng Long
Abstract: Deep models have recently emerged as a promising tool to solve partial differential equations (PDEs), known as neural PDE solvers. While neural solvers trained from either simulation data or physics-informed loss can solve the PDEs reasonably well, they are mainly restricted to a specific set of PDEs, e.g. a certain equation or a finite set of coefficients. This bottleneck limits the generalizability of neural solvers, which is widely recognized as its major advantage over numerical solvers. In this paper, we present the Universal PDE solver (Unisolver) capable of solving a wide scope of PDEs by leveraging a Transformer pre-trained on diverse data and conditioned on diverse PDEs. Instead of simply scaling up data and parameters, Unisolver stems from the theoretical analysis of the PDE-solving process. Our key finding is that a PDE solution is fundamentally under the control of a series of PDE components, e.g. equation symbols, coefficients, and initial and boundary conditions. Inspired by the mathematical structure of PDEs, we define a complete set of PDE components and correspondingly embed them as domain-wise (e.g. equation symbols) and point-wise (e.g. boundaries) conditions for Transformer PDE solvers. Integrating physical insights with recent Transformer advances, Unisolver achieves consistent state-of-the-art results on three challenging large-scale benchmarks, showing impressive gains and endowing favorable generalizability and scalability.
Authors: Haichao Sha, Yang Cao, Yong Liu, Yuncheng Wu, Ruixuan Liu, Hong Chen
Abstract: Differentially Private Stochastic Gradient Descent (DPSGD) is widely utilized to preserve training data privacy in deep learning, which first clips the gradients to a predefined norm and then injects calibrated noise into the training procedure. Existing DPSGD works typically assume the gradients follow sub-Gaussian distributions and design various clipping mechanisms to optimize training performance. However, recent studies have shown that the gradients in deep learning exhibit a heavy-tail phenomenon, that is, the tails of the gradient have infinite variance, which may lead to excessive clipping loss to the gradients with existing DPSGD mechanisms. To address this problem, we propose a novel approach, Discriminative Clipping~(DC)-DPSGD, with two key designs. First, we introduce a subspace identification technique to distinguish between body and tail gradients. Second, we present a discriminative clipping mechanism that applies different clipping thresholds for body and tail gradients to reduce the clipping loss. Under the non-convex condition, \ourtech{} reduces the empirical gradient norm from {${\mathbb{O}\left(\log^{\max(0,\theta-1)}(T/\delta)\log^{2\theta}(\sqrt{T})\right)}$} to {${\mathbb{O}\left(\log(\sqrt{T})\right)}$} with heavy-tailed index $\theta\geq 1/2$, iterations $T$, and arbitrary probability $\delta$. Extensive experiments on four real-world datasets demonstrate that our approach outperforms three baselines by up to 9.72\% in terms of accuracy.
Authors: Biqing Qi, Junqi Gao, Kaiyan Zhang, Dong Li, Jianxing Liu, Ligang Wu, Bowen Zhou
Abstract: Despite the promising performance of state space models (SSMs) in long sequence modeling, limitations still exist. Advanced SSMs like S5 and S6 (Mamba) in addressing non-uniform sampling, their recursive structures impede efficient SSM computation via convolution. To overcome compatibility limitations in parallel convolutional computation, this paper proposes a novel non-recursive non-uniform sample processing strategy. Theoretical analysis of SSMs through the lens of Event-Triggered Control (ETC) theory reveals the Non-Stable State (NSS) problem, where deviations from sampling point requirements lead to error transmission and accumulation, causing the divergence of the SSM's hidden state. Our analysis further reveals that adjustments of input sequences with early memories can mitigate the NSS problem, achieving Sampling Step Adaptation (SSA). Building on this insight, we introduce a simple yet effective plug-and-play mechanism, State Memory Replay (SMR), which utilizes learnable memories to adjust the current state with multi-step information for generalization at sampling points different from those in the training data. This enables SSMs to stably model varying sampling points. Experiments on long-range modeling tasks in autoregressive language modeling and Long Range Arena demonstrate the general effectiveness of the SMR mechanism for a series of SSM models.
Authors: Mucong Ding, Yuancheng Xu, Tahseen Rabbani, Xiaoyu Liu, Brian Gravelle, Teresa Ranadive, Tai-Ching Tuan, Furong Huang
Abstract: Dataset condensation can be used to reduce the computational cost of training multiple models on a large dataset by condensing the training dataset into a small synthetic set. State-of-the-art approaches rely on matching the model gradients between the real and synthetic data. However, there is no theoretical guarantee of the generalizability of the condensed data: data condensation often generalizes poorly across hyperparameters/architectures in practice. This paper considers a different condensation objective specifically geared toward hyperparameter search. We aim to generate a synthetic validation dataset so that the validation-performance rankings of the models, with different hyperparameters, on the condensed and original datasets are comparable. We propose a novel hyperparameter-calibrated dataset condensation (HCDC) algorithm, which obtains the synthetic validation dataset by matching the hyperparameter gradients computed via implicit differentiation and efficient inverse Hessian approximation. Experiments demonstrate that the proposed framework effectively maintains the validation-performance rankings of models and speeds up hyperparameter/architecture search for tasks on both images and graphs.
Authors: Giovanni De Toni, Nastaran Okati, Suhas Thejaswi, Eleni Straitouri, Manuel Gomez-Rodriguez
Abstract: Decision support systems based on prediction sets have proven to be effective at helping human experts solve classification tasks. Rather than providing single-label predictions, these systems provide sets of label predictions constructed using conformal prediction, namely prediction sets, and ask human experts to predict label values from these sets. In this paper, we first show that the prediction sets constructed using conformal prediction are, in general, suboptimal in terms of average accuracy. Then, we show that the problem of finding the optimal prediction sets under which the human experts achieve the highest average accuracy is NP-hard. More strongly, unless P = NP, we show that the problem is hard to approximate to any factor less than the size of the label set. However, we introduce a simple and efficient greedy algorithm that, for a large class of expert models and non-conformity scores, is guaranteed to find prediction sets that provably offer equal or greater performance than those constructed using conformal prediction. Further, using a simulation study with both synthetic and real expert predictions, we demonstrate that, in practice, our greedy algorithm finds near-optimal prediction sets offering greater performance than conformal prediction.
Authors: David Boetius, Stefan Leue, Tobias Sutter
Abstract: Probabilistic verification of neural networks is concerned with formally analysing the output distribution of a neural network under a probability distribution of the inputs. Examples of probabilistic verification include verifying the demographic parity fairness notion or quantifying the safety of a neural network. We present a new algorithm for the probabilistic verification of neural networks based on an algorithm for computing and iteratively refining lower and upper bounds on probabilities over the outputs of a neural network. By applying state-of-the-art bound propagation and branch and bound techniques from non-probabilistic neural network verification, our algorithm significantly outpaces existing probabilistic verification algorithms, reducing solving times for various benchmarks from the literature from tens of minutes to tens of seconds. Furthermore, our algorithm compares favourably even to dedicated algorithms for restricted subsets of probabilistic verification. We complement our empirical evaluation with a theoretical analysis, proving that our algorithm is sound and, under mildly restrictive conditions, also complete when using a suitable set of heuristics.
Authors: Marcelo Matheus Gauy, Larissa Cristina Berti, Arnaldo C\^andido Jr, Augusto Camargo Neto, Alfredo Goldman, Anna Sara Shafferman Levin, Marcus Martins, Beatriz Raposo de Medeiros, Marcelo Queiroz, Ester Cerdeira Sabino, Flaviane Romani Fernandes Svartman, Marcelo Finger
Abstract: This work investigates Artificial Intelligence (AI) systems that detect respiratory insufficiency (RI) by analyzing speech audios, thus treating speech as a RI biomarker. Previous works collected RI data (P1) from COVID-19 patients during the first phase of the pandemic and trained modern AI models, such as CNNs and Transformers, which achieved $96.5\%$ accuracy, showing the feasibility of RI detection via AI. Here, we collect RI patient data (P2) with several causes besides COVID-19, aiming at extending AI-based RI detection. We also collected control data from hospital patients without RI. We show that the considered models, when trained on P1, do not generalize to P2, indicating that COVID-19 RI has features that may not be found in all RI types.
Authors: Florent Forest, Katharina Rombach, Olga Fink
Abstract: Deep learning approaches have recently been extensively explored for the prognostics of industrial assets. However, they still suffer from a lack of interpretability, which hinders their adoption in safety-critical applications. To improve their trustworthiness, explainable AI (XAI) techniques have been applied in prognostics, primarily to quantify the importance of input variables for predicting the remaining useful life (RUL) using post-hoc attribution methods. In this work, we propose the application of Concept Bottleneck Models (CBMs), a family of inherently interpretable neural network architectures based on concept explanations, to the task of RUL prediction. Unlike attribution methods, which explain decisions in terms of low-level input features, concepts represent high-level information that is easily understandable by users. Moreover, once verified in actual applications, CBMs enable domain experts to intervene on the concept activations at test-time. We propose using the different degradation modes of an asset as intermediate concepts. Our case studies on the New Commercial Modular AeroPropulsion System Simulation (N-CMAPSS) aircraft engine dataset for RUL prediction demonstrate that the performance of CBMs can be on par or superior to black-box models, while being more interpretable, even when the available labeled concepts are limited. Code available at \href{https://github.com/EPFL-IMOS/concept-prognostics/}{\url{github.com/EPFL-IMOS/concept-prognostics/}}.
Authors: Zhenfeng Tu, Santiago Aranguri, Arthur Jacot
Abstract: The training dynamics of linear networks are well studied in two distinct setups: the lazy regime and balanced/active regime, depending on the initialization and width of the network. We provide a surprisingly simple unyfing formula for the evolution of the learned matrix that contains as special cases both lazy and balanced regimes but also a mixed regime in between the two. In the mixed regime, a part of the network is lazy while the other is balanced. More precisely the network is lazy along singular values that are below a certain threshold and balanced along those that are above the same threshold. At initialization, all singular values are lazy, allowing for the network to align itself with the task, so that later in time, when some of the singular value cross the threshold and become active they will converge rapidly (convergence in the balanced regime is notoriously difficult in the absence of alignment). The mixed regime is the `best of both worlds': it converges from any random initialization (in contrast to balanced dynamics which require special initialization), and has a low rank bias (absent in the lazy dynamics). This allows us to prove an almost complete phase diagram of training behavior as a function of the variance at initialization and the width, for a MSE training task.
Authors: Nguyen Phuc Tran, Duy Thanh Tran, Thi Thuy Nga Duong
Abstract: In recent years, a study by environmental organizations in the world and Vietnam shows that weather change is quite complex. global warming has become a serious problem in the modern world, which is a concern for scientists. last century, it was difficult to forecast the weather due to missing weather monitoring stations and technological limitations. this made it hard to collect data for building predictive models to make accurate simulations. in Vietnam, research on weather forecast models is a recent development, having only begun around 2000. along with advancements in computer science, mathematical models are being built and applied with machine learning techniques to create more accurate and reliable predictive models. this article will summarize the research and solutions for applying recurrent neural networks to forecast urban temperatures.
Authors: Meng Ding, Kaiyi Ji, Di Wang, Jinhui Xu
Abstract: Continual learning, focused on sequentially learning multiple tasks, has gained significant attention recently. Despite the tremendous progress made in the past, the theoretical understanding, especially factors contributing to catastrophic forgetting, remains relatively unexplored. In this paper, we provide a general theoretical analysis of forgetting in the linear regression model via Stochastic Gradient Descent (SGD) applicable to both underparameterized and overparameterized regimes. Our theoretical framework reveals some interesting insights into the intricate relationship between task sequence and algorithmic parameters, an aspect not fully captured in previous studies due to their restrictive assumptions. Specifically, we demonstrate that, given a sufficiently large data size, the arrangement of tasks in a sequence, where tasks with larger eigenvalues in their population data covariance matrices are trained later, tends to result in increased forgetting. Additionally, our findings highlight that an appropriate choice of step size will help mitigate forgetting in both underparameterized and overparameterized settings. To validate our theoretical analysis, we conducted simulation experiments on both linear regression models and Deep Neural Networks (DNNs). Results from these simulations substantiate our theoretical findings.
Authors: Klaudia Ba{\l}azy, Mohammadreza Banaei, Karl Aberer, Jacek Tabor
Abstract: The recent trend in scaling language models has led to a growing demand for parameter-efficient tuning (PEFT) methods such as LoRA (Low-Rank Adaptation). LoRA consistently matches or surpasses the full fine-tuning baseline with fewer parameters. However, handling numerous task-specific or user-specific LoRA modules on top of a base model still presents significant storage challenges. To address this, we introduce LoRA-XS (Low-Rank Adaptation with eXtremely Small number of parameters), a novel approach leveraging Singular Value Decomposition (SVD) for parameter-efficient fine-tuning. LoRA-XS introduces a small r x r weight matrix between frozen LoRA matrices, which are constructed by SVD of the original weight matrix. Training only r x r weight matrices ensures independence from model dimensions, enabling more parameter-efficient fine-tuning, especially for larger models. LoRA-XS achieves a remarkable reduction of trainable parameters by over 100x in 7B models compared to LoRA. Our benchmarking across various scales, including GLUE, GSM8k, and MATH benchmarks, shows that our approach outperforms LoRA and recent state-of-the-art approaches like VeRA in terms of parameter efficiency while maintaining competitive performance.
Authors: Ju-Seung Byun, Andrew Perrault
Abstract: Reinforcement learning (RL) training is inherently unstable due to factors such as moving targets and high gradient variance. Reinforcement Learning from Human Feedback (RLHF) and Reinforcement Learning from AI Feedback (RLAIF) can introduce additional difficulty. Differing preferences can complicate the alignment process, and prediction errors in a trained reward model can become more severe as the LLM generates unseen outputs. To enhance training robustness, RL has adopted techniques from supervised learning, such as ensembles and layer normalization. In this work, we improve the stability of RL training by adapting the reverse cross entropy (RCE) from supervised learning for noisy data to define a symmetric RL loss. We demonstrate performance improvements across various tasks and scales. We conduct experiments in discrete action tasks (Atari games) and continuous action space tasks (MuJoCo benchmark and Box2D) using Symmetric A2C (SA2C) and Symmetric PPO (SPPO), with and without added noise with especially notable performance in SPPO across different hyperparameters. Furthermore, we validate the benefits of the symmetric RL loss when using SPPO for large language models through improved performance in RLHF tasks, such as IMDB positive sentiment sentiment and TL;DR summarization tasks.
Authors: Sergio Rozada, Antonio G. Marques
Abstract: Most methods in reinforcement learning use a Policy Gradient (PG) approach to learn a parametric stochastic policy that maps states to actions. The standard approach is to implement such a mapping via a neural network (NN) whose parameters are optimized using stochastic gradient descent. However, PG methods are prone to large policy updates that can render learning inefficient. Trust region algorithms, like Trust Region Policy Optimization (TRPO), constrain the policy update step, ensuring monotonic improvements. This paper introduces low-rank matrix-based models as an efficient alternative for estimating the parameters of TRPO algorithms. By gathering the stochastic policy's parameters into a matrix and applying matrix-completion techniques, we promote and enforce low rank. Our numerical studies demonstrate that low-rank matrix-based policy models effectively reduce both computational and sample complexities compared to NN models, while maintaining comparable aggregated rewards.
Authors: Sergio Rozada, Antonio G. Marques
Abstract: Estimating a policy that maps states to actions is a central problem in reinforcement learning. Traditionally, policies are inferred from the so called value functions (VFs), but exact VF computation suffers from the curse of dimensionality. Policy gradient (PG) methods bypass this by learning directly a parametric stochastic policy. Typically, the parameters of the policy are estimated using neural networks (NNs) tuned via stochastic gradient descent. However, finding adequate NN architectures can be challenging, and convergence issues are common as well. In this paper, we put forth low-rank matrix-based models to estimate efficiently the parameters of PG algorithms. We collect the parameters of the stochastic policy into a matrix, and then, we leverage matrix-completion techniques to promote (enforce) low rank. We demonstrate via numerical studies how low-rank matrix-based policy models reduce the computational and sample complexities relative to NN models, while achieving a similar aggregated reward.
Authors: Wenxiao Xiao, Hongfu Liu
Abstract: Active learning strategically selects informative unlabeled data points and queries their ground truth labels for model training. The prevailing assumption underlying this machine learning paradigm is that acquiring these ground truth labels will optimally enhance model performance. However, this assumption may not always hold true or maximize learning capacity, particularly considering the costly labor annotations required for ground truth labels. In contrast to traditional ground truth labeling, this paper proposes salutary labeling, which automatically assigns the most beneficial labels to the most informative samples without human annotation. Specifically, we utilize the influence function, a tool for estimating sample influence, to select newly added samples and assign their salutary labels by choosing the category that maximizes their positive influence. This process eliminates the need for human annotation. Extensive experiments conducted on nine benchmark datasets demonstrate the superior performance of our salutary labeling approach over traditional active learning strategies. Additionally, we provide several in-depth explorations and practical applications of large language model (LLM) fine-tuning.
Authors: Sergio Rozada, Antonio G. Marques
Abstract: The goal of reinforcement learning is estimating a policy that maps states to actions and maximizes the cumulative reward of a Markov Decision Process (MDP). This is oftentimes achieved by estimating first the optimal (reward) value function (VF) associated with each state-action pair. When the MDP has an infinite horizon, the optimal VFs and policies are stationary under mild conditions. However, in finite-horizon MDPs, the VFs (hence, the policies) vary with time. This poses a challenge since the number of VFs to estimate grows not only with the size of the state-action space but also with the time horizon. This paper proposes a non-parametric low-rank stochastic algorithm to approximate the VFs of finite-horizon MDPs. First, we represent the (unknown) VFs as a multi-dimensional array, or tensor, where time is one of the dimensions. Then, we use rewards sampled from the MDP to estimate the optimal VFs. More precisely, we use the (truncated) PARAFAC decomposition to design an online low-rank algorithm that recovers the entries of the tensor of VFs. The size of the low-rank PARAFAC model grows additively with respect to each of its dimensions, rendering our approach efficient, as demonstrated via numerical experiments.
Authors: Xiaoou Cheng, Jonathan Weare
Abstract: We quantify the efficiency of temporal difference (TD) learning over the direct, or Monte Carlo (MC), estimator for policy evaluation in reinforcement learning, with an emphasis on estimation of quantities related to rare events. Policy evaluation is complicated in the rare event setting by the long timescale of the event and by the need for \emph{relative accuracy} in estimates of very small values. Specifically, we focus on least-squares TD (LSTD) prediction for finite state Markov chains, and show that LSTD can achieve relative accuracy far more efficiently than MC. We prove a central limit theorem for the LSTD estimator and upper bound the \emph{relative asymptotic variance} by simple quantities characterizing the connectivity of states relative to the transition probabilities between them. Using this bound, we show that, even when both the timescale of the rare event and the relative accuracy of the MC estimator are exponentially large in the number of states, LSTD maintains a fixed level of relative accuracy with a total number of observed transitions of the Markov chain that is only \emph{polynomially} large in the number of states.
Authors: Patryk Wielopolski, Oleksii Furman, Jerzy Stefanowski, Maciej Zi\k{e}ba
Abstract: We present PPCEF, a novel method for generating probabilistically plausible counterfactual explanations (CFs). PPCEF advances beyond existing methods by combining a probabilistic formulation that leverages the data distribution with the optimization of plausibility within a unified framework. Compared to reference approaches, our method enforces plausibility by directly optimizing the explicit density function without assuming a particular family of parametrized distributions. This ensures CFs are not only valid (i.e., achieve class change) but also align with the underlying data's probability density. For that purpose, our approach leverages normalizing flows as powerful density estimators to capture the complex high-dimensional data distribution. Furthermore, we introduce a novel loss that balances the trade-off between achieving class change and maintaining closeness to the original instance while also incorporating a probabilistic plausibility term. PPCEF's unconstrained formulation allows for efficient gradient-based optimization with batch processing, leading to orders of magnitude faster computation compared to prior methods. Moreover, the unconstrained formulation of PPCEF allows for the seamless integration of future constraints tailored to specific counterfactual properties. Finally, extensive evaluations demonstrate PPCEF's superiority in generating high-quality, probabilistically plausible counterfactual explanations in high-dimensional tabular settings. This makes PPCEF a powerful tool for not only interpreting complex machine learning models but also for improving fairness, accountability, and trust in AI systems.
Authors: Patryk Wielopolski, Oleksii Furman, Jerzy Stefanowski, Maciej Zi\k{e}ba
Abstract: Growing regulatory and societal pressures demand increased transparency in AI, particularly in understanding the decisions made by complex machine learning models. Counterfactual Explanations (CFs) have emerged as a promising technique within Explainable AI (xAI), offering insights into individual model predictions. However, to understand the systemic biases and disparate impacts of AI models, it is crucial to move beyond local CFs and embrace global explanations, which offer a~holistic view across diverse scenarios and populations. Unfortunately, generating Global Counterfactual Explanations (GCEs) faces challenges in computational complexity, defining the scope of "global," and ensuring the explanations are both globally representative and locally plausible. We introduce a novel unified approach for generating Local, Group-wise, and Global Counterfactual Explanations for differentiable classification models via gradient-based optimization to address these challenges. This framework aims to bridge the gap between individual and systemic insights, enabling a deeper understanding of model decisions and their potential impact on diverse populations. Our approach further innovates by incorporating a probabilistic plausibility criterion, enhancing actionability and trustworthiness. By offering a cohesive solution to the optimization and plausibility challenges in GCEs, our work significantly advances the interpretability and accountability of AI models, marking a step forward in the pursuit of transparent AI.
Authors: Xinting Huang, Madhur Panwar, Navin Goyal, Michael Hahn
Abstract: The inner workings of neural networks can be better understood if we can fully decipher the information encoded in neural activations. In this paper, we argue that this information is embodied by the subset of inputs that give rise to similar activations. Computing such subsets is nontrivial as the input space is exponentially large. We propose InversionView, which allows us to practically inspect this subset by sampling from a trained decoder model conditioned on activations. This helps uncover the information content of activation vectors, and facilitates understanding of the algorithms implemented by transformer models. We present three case studies where we investigate models ranging from small transformers to GPT-2. In these studies, we demonstrate the characteristics of our method, show the distinctive advantages it offers, and provide causally verified circuits.
Authors: Najwa Laabid, Severi Rissanen, Markus Heinonen, Arno Solin, Vikas Garg
Abstract: Retrosynthesis, the task of identifying precursors for a given molecule, can be naturally framed as a conditional graph generation task. Diffusion models are a particularly promising modelling approach, enabling post-hoc conditioning and trading off quality for speed during generation. We show mathematically that permutation equivariant denoisers severely limit the expressiveness of graph diffusion models and thus their adaptation to retrosynthesis. To address this limitation, we relax the equivariance requirement such that it only applies to aligned permutations of the conditioning and the generated graphs obtained through atom mapping. Our new denoiser achieves the highest top-$1$ accuracy ($54.7$\%) across template-free and template-based methods on USPTO-50k. We also demonstrate the ability for flexible post-training conditioning and good sample quality with small diffusion step counts, highlighting the potential for interactive applications and additional controls for multi-step planning.
Authors: Cory Efird, Alex Murphy, Joel Zylberberg, Alona Fyshe
Abstract: Prior work has offered evidence for functional localization in the brain; different anatomical regions preferentially activate for certain types of visual input. For example, the fusiform face area preferentially activates for visual stimuli that include a face. However, the spectrum of visual semantics is extensive, and only a few semantically-tuned patches of cortex have so far been identified in the human brain. Using a multimodal (natural language and image) neural network architecture (CLIP) we train a highly accurate contrastive model that maps brain responses during naturalistic image viewing to CLIP embeddings. We then use a novel adaptation of the DBSCAN clustering algorithm to cluster the parameters of these participant-specific contrastive models. This reveals what we call Shared Decodable Concepts (SDCs): clusters in CLIP space that are decodable from common sets of voxels across multiple participants. Examining the images most and least associated with each SDC cluster gives us additional insight into the semantic properties of each SDC. We note SDCs for previously reported visual features (e.g. orientation tuning in early visual cortex) as well as visual semantic concepts such as faces, places and bodies. In cases where our method finds multiple clusters for a visuo-semantic concept, the least associated images allow us to dissociate between confounding factors. For example, we discovered two clusters of food images, one driven by color, the other by shape. We also uncover previously unreported areas such as regions of extrastriate body area (EBA) tuned for legs/hands and sensitivity to numerosity in right intraparietal sulcus, and more. Thus, our contrastive-learning methodology better characterizes new and existing visuo-semantic representations in the brain by leveraging multimodal neural network representations and a novel adaptation of clustering algorithms.
Authors: Lukasz Sztukiewicz, Jack Henry Good, Artur Dubrawski
Abstract: In the real world, data is often noisy, affecting not only the quality of features but also the accuracy of labels. Current research on mitigating label errors stems primarily from advances in deep learning, and a gap exists in exploring interpretable models, particularly those rooted in decision trees. In this study, we investigate whether ideas from deep learning loss design can be applied to improve the robustness of decision trees. In particular, we show that loss correction and symmetric losses, both standard approaches, are not effective. We argue that other directions need to be explored to improve the robustness of decision trees to label noise.
Authors: Matan Goren, Eran Treister
Abstract: Full Waveform Inversion (FWI) is an inverse problem for estimating the wave velocity distribution in a given domain, based on observed data on the boundaries. The inversion is computationally demanding because we are required to solve multiple forward problems, either in time or frequency domains, to simulate data that are then iteratively fitted to the observed data. We consider FWI in the frequency domain, where the Helmholtz equation is used as a forward model, and its repeated solution is the main computational bottleneck of the inversion process. To ease this cost, we integrate a learning process of an encoder-solver preconditioner that is based on convolutional neural networks (CNNs). The encoder-solver is trained to effectively precondition the discretized Helmholtz operator given velocity medium parameters. Then, by re-training the CNN between the iterations of the optimization process, the encoder-solver is adapted to the iteratively evolving velocity medium as part of the inversion. Without retraining, the performance of the solver deteriorates as the medium changes. Using our light retraining procedures, we obtain the forward simulations effectively throughout the process. We demonstrate our approach to solving FWI problems using 2D geophysical models with high-frequency data.
Authors: Mohammad Mahdi Maheri, Sandra Siby, Ali Shahin Shamsabadi, Sina Abdollahi, Anastasia Borovykh, Hamed Haddadi
Abstract: Personalized learning is a proposed approach to address the problem of data heterogeneity in collaborative machine learning. In a decentralized setting, the two main challenges of personalization are client clustering and data privacy. In this paper, we address these challenges by developing P4 (Personalized Private Peer-to-Peer) a method that ensures that each client receives a personalized model while maintaining differential privacy guarantee of each client's local dataset during and after the training. Our approach includes the design of a lightweight algorithm to identify similar clients and group them in a private, peer-to-peer (P2P) manner. Once grouped, we develop differentially-private knowledge distillation for clients to co-train with minimal impact on accuracy. We evaluate our proposed method on three benchmark datasets (FEMNIST or Federated EMNIST, CIFAR-10 and CIFAR-100) and two different neural network architectures (Linear and CNN-based networks) across a range of privacy parameters. The results demonstrate the potential of P4, as it outperforms the state-of-the-art of differential private P2P by up to 40 percent in terms of accuracy. We also show the practicality of P4 by implementing it on resource constrained devices, and validating that it has minimal overhead, e.g., about 7 seconds to run collaborative training between two clients.
Authors: Jason Li
Abstract: Recent research (arXiv:2310.11453, arXiv:2402.17764) has proposed binary and ternary transformer networks as a way to significantly reduce memory and improve inference speed in Large Language Models (LLMs) while maintaining accuracy. In this work, we apply techniques from mechanistic interpretability to investigate whether such networks learn distinctly different or similar algorithms when compared to full-precision transformer networks. In particular, we reverse engineer the algorithms learned for the toy problem of modular addition where we find that binary and ternary networks learn similar algorithms as full precision networks. This provides evidence against the possibility of using binary and ternary networks as a more interpretable alternative in the LLM setting.
Authors: Allen Nie, Yash Chandak, Christina J. Yuan, Anirudhan Badrinath, Yannis Flet-Berliac, Emma Brunskil
Abstract: Offline policy evaluation (OPE) allows us to evaluate and estimate a new sequential decision-making policy's performance by leveraging historical interaction data collected from other policies. Evaluating a new policy online without a confident estimate of its performance can lead to costly, unsafe, or hazardous outcomes, especially in education and healthcare. Several OPE estimators have been proposed in the last decade, many of which have hyperparameters and require training. Unfortunately, choosing the best OPE algorithm for each task and domain is still unclear. In this paper, we propose a new algorithm that adaptively blends a set of OPE estimators given a dataset without relying on an explicit selection using a statistical procedure. We prove that our estimator is consistent and satisfies several desirable properties for policy evaluation. Additionally, we demonstrate that when compared to alternative approaches, our estimator can be used to select higher-performing policies in healthcare and robotics. Our work contributes to improving ease of use for a general-purpose, estimator-agnostic, off-policy evaluation framework for offline RL.
Authors: Yanbing Bai, Xinyi Wu, Lai Xu, Jihan Pei, Erick Mas, Shunichi Koshimura
Abstract: With the rapid development of earth observation technology, we have entered an era of massively available satellite remote-sensing data. However, a large amount of satellite remote sensing data lacks a label or the label cost is too high to hinder the potential of AI technology mining satellite data. Especially in such an emergency response scenario that uses satellite data to evaluate the degree of disaster damage. Disaster damage assessment encountered bottlenecks due to excessive focus on the damage of a certain building in a specific geographical space or a certain area on a larger scale. In fact, in the early days of disaster emergency response, government departments were more concerned about the overall damage rate of the disaster area instead of single-building damage, because this helps the government decide the level of emergency response. We present an innovative algorithm that constructs Neyman stratified random sampling trees for binary classification and extends this approach to multiclass problems. Through extensive experimentation on various datasets and model structures, our findings demonstrate that our method surpasses both passive and conventional active learning techniques in terms of class rate estimation and model enhancement with only 30\%-60\% of the annotation cost of simple sampling. It effectively addresses the 'sampling bias' challenge in traditional active learning strategies and mitigates the 'cold start' dilemma. The efficacy of our approach is further substantiated through application to disaster evaluation tasks using Xview2 Satellite imagery, showcasing its practical utility in real-world contexts.
Authors: Nan Li, Haiyang Yu, Ping Yi
Abstract: Deep Neural Networks (DNNs) are known to be vulnerable to backdoor attacks, posing concerning threats to their reliable deployment. Recent research reveals that backdoors can be erased from infected DNNs by pruning a specific group of neurons, while how to effectively identify and remove these backdoor-associated neurons remains an open challenge. Most of the existing defense methods rely on defined rules and focus on neuron's local properties, ignoring the exploration and optimization of pruning policies. To address this gap, we propose an Optimized Neuron Pruning (ONP) method combined with Graph Neural Network (GNN) and Reinforcement Learning (RL) to repair backdoor models. Specifically, ONP first models the target DNN as graphs based on neuron connectivity, and then uses GNN-based RL agents to learn graph embeddings and find a suitable pruning policy. To the best of our knowledge, this is the first attempt to employ GNN and RL for optimizing pruning policies in the field of backdoor defense. Experiments show, with a small amount of clean data, ONP can effectively prune the backdoor neurons implanted by a set of backdoor attacks at the cost of negligible performance degradation, achieving a new state-of-the-art performance for backdoor mitigation.
Authors: Nan Li, Haoyu Jiang, Ping Yi
Abstract: Deep Neural Networks (DNNs) are known to be vulnerable to backdoor attacks, posing concerning threats to their reliable deployment. Recent research reveals that backdoors can be erased from infected DNNs by pruning a specific group of neurons, while how to effectively identify and remove these backdoor-associated neurons remains an open challenge. In this paper, we investigate the correlation between backdoor behavior and neuron magnitude, and find that backdoor neurons deviate from the magnitude-saliency correlation of the model. The deviation inspires us to propose a Magnitude-based Neuron Pruning (MNP) method to detect and prune backdoor neurons. Specifically, MNP uses three magnitude-guided objective functions to manipulate the magnitude-saliency correlation of backdoor neurons, thus achieving the purpose of exposing backdoor behavior, eliminating backdoor neurons and preserving clean neurons, respectively. Experiments show our pruning strategy achieves state-of-the-art backdoor defense performance against a variety of backdoor attacks with a limited amount of clean data, demonstrating the crucial role of magnitude for guiding backdoor defenses.
Authors: Hao Di, Haishan Ye, Yueling Zhang, Xiangyu Chang, Guang Dai, Ivor W. Tsang
Abstract: Variance reduction techniques are designed to decrease the sampling variance, thereby accelerating convergence rates of first-order (FO) and zeroth-order (ZO) optimization methods. However, in composite optimization problems, ZO methods encounter an additional variance called the coordinate-wise variance, which stems from the random gradient estimation. To reduce this variance, prior works require estimating all partial derivatives, essentially approximating FO information. This approach demands O(d) function evaluations (d is the dimension size), which incurs substantial computational costs and is prohibitive in high-dimensional scenarios. This paper proposes the Zeroth-order Proximal Double Variance Reduction (ZPDVR) method, which utilizes the averaging trick to reduce both sampling and coordinate-wise variances. Compared to prior methods, ZPDVR relies solely on random gradient estimates, calls the stochastic zeroth-order oracle (SZO) in expectation $\mathcal{O}(1)$ times per iteration, and achieves the optimal $\mathcal{O}(d(n + \kappa)\log (\frac{1}{\epsilon}))$ SZO query complexity in the strongly convex and smooth setting, where $\kappa$ represents the condition number and $\epsilon$ is the desired accuracy. Empirical results validate ZPDVR's linear convergence and demonstrate its superior performance over other related methods.
Authors: Rahul Thapa, Bryan He, Magnus Ruud Kjaer, Hyatt Moore, Gauri Ganjoo, Emmanuel Mignot, James Zou
Abstract: Sleep is a complex physiological process evaluated through various modalities recording electrical brain, cardiac, and respiratory activities. We curate a large polysomnography dataset from over 14,000 participants comprising over 100,000 hours of multi-modal sleep recordings. Leveraging this extensive dataset, we developed SleepFM, the first multi-modal foundation model for sleep analysis. We show that a novel leave-one-out approach for contrastive learning significantly improves downstream task performance compared to representations from standard pairwise contrastive learning. A logistic regression model trained on SleepFM's learned embeddings outperforms an end-to-end trained convolutional neural network (CNN) on sleep stage classification (macro AUROC 0.88 vs 0.72 and macro AUPRC 0.72 vs 0.48) and sleep disordered breathing detection (AUROC 0.85 vs 0.69 and AUPRC 0.77 vs 0.61). Notably, the learned embeddings achieve 48% top-1 average accuracy in retrieving the corresponding recording clips of other modalities from 90,000 candidates. This work demonstrates the value of holistic multi-modal sleep modeling to fully capture the richness of sleep recordings. SleepFM is open source and available at https://github.com/rthapa84/sleepfm-codebase.
Authors: Robert Wu, Vardan Papyan
Abstract: Neural collapse ($\mathcal{NC}$) is a phenomenon observed in classification tasks where top-layer representations collapse into their class means, which become equinorm, equiangular and aligned with the classifiers. These behaviors -- associated with generalization and robustness -- would manifest under specific conditions: models are trained towards zero loss, with noise-free labels belonging to balanced classes, which do not outnumber the model's hidden dimension. Recent studies have explored $\mathcal{NC}$ in the absence of one or more of these conditions to extend and capitalize on the associated benefits of ideal geometries. Language modeling presents a curious frontier, as \textit{training by token prediction} constitutes a classification task where none of the conditions exist: the vocabulary is imbalanced and exceeds the embedding dimension; different tokens might correspond to similar contextual embeddings; and large language models (LLMs) in particular are typically only trained for a few epochs. This paper empirically investigates the impact of scaling the architectures and training of causal language models (CLMs) on their progression towards $\mathcal{NC}$. We find that $\mathcal{NC}$ properties that develop with scaling are linked to generalization. Moreover, there is evidence of some relationship between $\mathcal{NC}$ and generalization independent of scale. Our work therefore underscores the generality of $\mathcal{NC}$ as it extends to the novel and more challenging setting of language modeling. Downstream, we seek to inspire further research on the phenomenon to deepen our understanding of LLMs -- and neural networks at large -- and improve existing architectures based on $\mathcal{NC}$-related properties.
Authors: Zhuonan Zheng, Yuanchen Bei, Sheng Zhou, Yao Ma, Ming Gu, HongJia XU, Chengyu Lai, Jiawei Chen, Jiajun Bu
Abstract: Graph Neural Networks (GNNs) have demonstrated strong performance in graph mining tasks due to their message-passing mechanism, which is aligned with the homophily assumption that adjacent nodes exhibit similar behaviors. However, in many real-world graphs, connected nodes may display contrasting behaviors, termed as heterophilous patterns, which has attracted increased interest in heterophilous GNNs (HTGNNs). Although the message-passing mechanism seems unsuitable for heterophilous graphs due to the propagation of class-irrelevant information, it is still widely used in many existing HTGNNs and consistently achieves notable success. This raises the question: why does message passing remain effective on heterophilous graphs? To answer this question, in this paper, we revisit the message-passing mechanisms in heterophilous graph neural networks and reformulate them into a unified heterophilious message-passing (HTMP) mechanism. Based on HTMP and empirical analysis, we reveal that the success of message passing in existing HTGNNs is attributed to implicitly enhancing the compatibility matrix among classes. Moreover, we argue that the full potential of the compatibility matrix is not completely achieved due to the existence of incomplete and noisy semantic neighborhoods in real-world heterophilous graphs. To bridge this gap, we introduce a new approach named CMGNN, which operates within the HTMP mechanism to explicitly leverage and improve the compatibility matrix. A thorough evaluation involving 10 benchmark datasets and comparative analysis against 13 well-established baselines highlights the superior performance of the HTMP mechanism and CMGNN method.
Authors: Xingyu Ding, Lianlei Shan, Guiqin Zhao, Meiqi Wu, Wenzhang Zhou, Wei Li
Abstract: Deep learning-based information processing consumes long time and requires huge computing resources, especially for dense prediction tasks which require an output for each pixel, like semantic segmentation and salient object detection. There are mainly two challenges for quantization of dense prediction tasks. Firstly, directly applying the upsampling operation that dense prediction tasks require is extremely crude and causes unacceptable accuracy reduction. Secondly, the complex structure of dense prediction networks means it is difficult to maintain a fast speed as well as a high accuracy when performing quantization. In this paper, we propose an effective upsampling method and an efficient attention computation strategy to transfer the success of the binary neural networks (BNN) from single prediction tasks to dense prediction tasks. Firstly, we design a simple and robust multi-branch parallel upsampling structure to achieve the high accuracy. Then we further optimize the attention method which plays an important role in segmentation but has huge computation complexity. Our attention method can reduce the computational complexity by a factor of one hundred times but retain the original effect. Experiments on Cityscapes, KITTI road, and ECSSD fully show the effectiveness of our work.
Authors: Huiping Zhuang, Di Fang, Kai Tong, Yuchen Liu, Ziqian Zeng, Xu Zhou, Cen Chen
Abstract: In the field of autonomous driving, even a meticulously trained model can encounter failures when faced with unfamiliar sceanrios. One of these scenarios can be formulated as an online continual learning (OCL) problem. That is, data come in an online fashion, and models are updated according to these streaming data. Two major OCL challenges are catastrophic forgetting and data imbalance. To address these challenges, in this paper, we propose an Analytic Exemplar-Free Online Continual Learning (AEF-OCL). The AEF-OCL leverages analytic continual learning principles and employs ridge regression as a classifier for features extracted by a large backbone network. It solves the OCL problem by recursively calculating the analytical solution, ensuring an equalization between the continual learning and its joint-learning counterpart, and works without the need to save any used samples (i.e., exemplar-free). Additionally, we introduce a Pseudo-Features Generator (PFG) module that recursively estimates the deviation of real features. The PFG generates offset pseudo-features following a normal distribution, thereby addressing the data imbalance issue. Experimental results demonstrate that despite being an exemplar-free strategy, our method outperforms various methods on the autonomous driving SODA10M dataset. Source code is available at https://github.com/ZHUANGHP/Analytic-continual-learning.
URLs: https://github.com/ZHUANGHP/Analytic-continual-learning.
Authors: Yuying Duan, Yijun Tian, Nitesh Chawla, Michael Lemmon
Abstract: Federated Learning (FL) is a distributed machine learning framework in which a set of local communities collaboratively learn a shared global model while retaining all training data locally within each community. Two notions of fairness have recently emerged as important issues for federated learning: group fairness and community fairness. Group fairness requires that a model's decisions do not favor any particular group based on a set of legally protected attributes such as race or gender. Community fairness requires that global models exhibit similar levels of performance (accuracy) across all collaborating communities. Both fairness concepts can coexist within an FL framework, but the existing literature has focused on either one concept or the other. This paper proposes and analyzes a post-processing fair federated learning (FFL) framework called post-FFL. Post-FFL uses a linear program to simultaneously enforce group and community fairness while maximizing the utility of the global model. Because Post-FFL is a post-processing approach, it can be used with existing FL training pipelines whose convergence properties are well understood. This paper uses post-FFL on real-world datasets to mimic how hospital networks, for example, use federated learning to deliver community health care. Theoretical results bound the accuracy lost when post-FFL enforces both notion of fairness. Experimental results illustrate that post-FFL simultaneously improves both group and community fairness in FL. Moreover, post-FFL outperforms the existing in-processing fair federated learning in terms of improving both notions of fairness, communication efficiency and computation cost.
Authors: Ignat Georgiev, Krishnan Srinivasan, Jie Xu, Eric Heiden, Animesh Garg
Abstract: Model-Free Reinforcement Learning~(MFRL), leveraging the policy gradient theorem, has demonstrated considerable success in continuous control tasks. However, these approaches are plagued by high gradient variance due to zeroth-order gradient estimation, resulting in suboptimal policies. Conversely, First-Order Model-Based Reinforcement Learning~(FO-MBRL) methods, employing differentiable simulation, provide gradients with reduced variance but are susceptible to sampling error in scenarios involving stiff dynamics, such as physical contact. This paper investigates the source of this error and introduces Adaptive Horizon Actor-Critic (AHAC), an FO-MBRL algorithm that reduces gradient error by adapting the model-based horizon to avoid stiff dynamics. Empirical findings reveal that AHAC outperforms MFRL baselines, attaining 40\% more reward across a set of locomotion tasks, and efficiently scaling to high-dimensional control environments with improved wall-clock-time efficiency.
Authors: Jian Qian, Haichen Hu, David Simchi-Levi
Abstract: Motivated by the recent discovery of a statistical and computational reduction from contextual bandits to offline regression (Simchi-Levi and Xu, 2021), we address the general (stochastic) Contextual Markov Decision Process (CMDP) problem with horizon H (as known as CMDP with H layers). In this paper, we introduce a reduction from CMDPs to offline density estimation under the realizability assumption, i.e., a model class M containing the true underlying CMDP is provided in advance. We develop an efficient, statistically near-optimal algorithm requiring only O(HlogT) calls to an offline density estimation algorithm (or oracle) across all T rounds of interaction. This number can be further reduced to O(HloglogT) if T is known in advance. Our results mark the first efficient and near-optimal reduction from CMDPs to offline density estimation without imposing any structural assumptions on the model class. A notable feature of our algorithm is the design of a layerwise exploration-exploitation tradeoff tailored to address the layerwise structure of CMDPs. Additionally, our algorithm is versatile and applicable to pure exploration tasks in reward-free reinforcement learning.
Authors: Yudong Wang, Damai Dai, Zhifang Sui
Abstract: Most work treats large language models as black boxes without in-depth understanding of their internal working mechanism. In order to explain the internal representations of LLMs, we propose a gradient-based metric to assess the activation level of model parameters. Based on this metric, we obtain three preliminary findings. (1) When the inputs are in the same domain, parameters in the shallow layers will be activated densely, which means a larger portion of parameters will have great impacts on the outputs. In contrast, parameters in the deep layers are activated sparsely. (2) When the inputs are across different domains, parameters in shallow layers exhibit higher similarity in the activation behavior than deep layers. (3) In deep layers, the similarity of the distributions of activated parameters is positively correlated to the empirical data relevance. Further, we develop three validation experiments to solidify these findings. (1) Firstly, starting from the first finding, we attempt to configure different prune ratios for different layers, and find this method can benefit model pruning. (2) Secondly, we find that a pruned model based on one calibration set can better handle tasks related to the calibration task than those not related, which validate the second finding. (3) Thirdly, Based on the STS-B and SICK benchmark, we find that two sentences with consistent semantics tend to share similar parameter activation patterns in deep layers, which aligns with our third finding. Our work sheds light on the behavior of parameter activation in LLMs, and we hope these findings will have the potential to inspire more practical applications.
Authors: Yuanle Mo, Xin Hong, Bowen Gao, Yinjun Jia, Yanyan Lan
Abstract: Protein-protein interactions are central mediators in many biological processes. Accurately predicting the effects of mutations on interactions is crucial for guiding the modulation of these interactions, thereby playing a significant role in therapeutic development and drug discovery. Mutations generally affect interactions hierarchically across three levels: mutated residues exhibit different sidechain conformations, which lead to changes in the backbone conformation, eventually affecting the binding affinity between proteins. However, existing methods typically focus only on sidechain-level interaction modeling, resulting in suboptimal predictions. In this work, we propose a self-supervised multi-level pre-training framework, ProMIM, to fully capture all three levels of interactions with well-designed pretraining objectives. Experiments show ProMIM outperforms all the baselines on the standard benchmark, especially on mutations where significant changes in backbone conformations may occur. In addition, leading results from zero-shot evaluations for SARS-CoV-2 mutational effect prediction and antibody optimization underscore the potential of ProMIM as a powerful next-generation tool for developing novel therapeutic approaches and new drugs.
Authors: Jinho Chang, Jong Chul Ye
Abstract: With the emergence of diffusion models as the frontline of generative models, many researchers have proposed molecule generation techniques using conditional diffusion models. However, due to the fundamental nature of a molecule, which carries highly entangled correlations within a small number of atoms and bonds, it becomes difficult for a model to connect raw data with the conditions when the conditions become more complex as natural language. To address this, here we present a novel latent diffusion model dubbed LDMol, which enables a natural text-conditioned molecule generation. Specifically, LDMol is composed of three building blocks: a molecule encoder that produces a chemically informative feature space, a natural language-conditioned latent diffusion model using a Diffusion Transformer (DiT), and an autoregressive decoder for molecule re. In particular, recognizing that multiple SMILES notations can represent the same molecule, we employ a contrastive learning strategy to extract the chemical informative feature space. LDMol not only beats the existing baselines on the text-to-molecule generation benchmark but is also capable of zero-shot inference with unseen scenarios. Furthermore, we show that LDMol can be applied to downstream tasks such as molecule-to-text retrieval and text-driven molecule editing, demonstrating its versatility as a diffusion model.
Authors: Tao Wang, Sylvia Herbert, Sicun Gao
Abstract: Policy gradient methods have enabled deep reinforcement learning (RL) to approach challenging continuous control problems, even when the underlying systems involve highly nonlinear dynamics that generate complex non-smooth optimization landscapes. We develop a rigorous framework for understanding how policy gradient methods mollify non-smooth optimization landscapes to enable effective policy search, as well as the downside of it: while making the objective function smoother and easier to optimize, the stochastic objective deviates further from the original problem. We demonstrate the equivalence between policy gradient methods and solving backward heat equations. Following the ill-posedness of backward heat equations from PDE theory, we present a fundamental challenge to the use of policy gradient under stochasticity. Moreover, we make the connection between this limitation and the uncertainty principle in harmonic analysis to understand the effects of exploration with stochastic policies in RL. We also provide experimental results to illustrate both the positive and negative aspects of mollification effects in practice.
Authors: Keith Burghardt, Daniel M. T. Fessler, Chyna Tang, Anne Pisor, Kristina Lerman
Abstract: Socio-linguistic indicators of text, such as emotion or sentiment, are often extracted using neural networks in order to better understand features of social media. One indicator that is often overlooked, however, is the presence of hazards within text. Recent psychological research suggests that statements about hazards are more believable than statements about benefits (a property known as negatively biased credulity), and that political liberals and conservatives differ in how often they share hazards. Here, we develop a new model to detect information concerning hazards, trained on a new collection of annotated X posts, as well as urban legends annotated in previous work. We show that not only does this model perform well (outperforming, e.g., zero-shot human annotator proxies, such as GPT-4) but that the hazard information it extracts is not strongly correlated with other indicators, namely moral outrage, sentiment, emotions, and threat words. (That said, consonant with expectations, hazard information does correlate positively with such emotions as fear, and negatively with emotions like joy.) We then apply this model to three datasets: X posts about COVID-19, X posts about the 2023 Hamas-Israel war, and a new expanded collection of urban legends. From these data, we uncover words associated with hazards unique to each dataset as well as differences in this language between groups of users, such as conservatives and liberals, which informs what these groups perceive as hazards. We further show that information about hazards peaks in frequency after major hazard events, and therefore acts as an automated indicator of such events. Finally, we find that information about hazards is especially prevalent in urban legends, which is consistent with previous work that finds that reports of hazards are more likely to be both believed and transmitted.
Authors: Xing Hu, Yuan Chen, Dawei Yang, Sifan Zhou, Zhihang Yuan, Jiangyong Yu, Chen Xu
Abstract: Post-training quantization (PTQ) serves as a potent technique to accelerate the inference of large language models (LLMs). Nonetheless, existing works still necessitate a considerable number of floating-point (FP) operations during inference, including additional quantization and de-quantization, as well as non-linear operators such as RMSNorm and Softmax. This limitation hinders the deployment of LLMs on the edge and cloud devices. In this paper, we identify the primary obstacle to integer-only quantization for LLMs lies in the large fluctuation of activations across channels and tokens in both linear and non-linear operations. To address this issue, we propose I-LLM, a novel integer-only fully-quantized PTQ framework tailored for LLMs. Specifically, (1) we develop Fully-Smooth Block-Reconstruction (FSBR) to aggressively smooth inter-channel variations of all activations and weights. (2) to alleviate degradation caused by inter-token variations, we introduce a novel approach called Dynamic Integer-only MatMul (DI-MatMul). This method enables dynamic quantization in full-integer matrix multiplication by dynamically quantizing the input and outputs with integer-only operations. (3) we design DI-ClippedSoftmax, DI-Exp, and DI-Normalization, which utilize bit shift to execute non-linear operators efficiently while maintaining accuracy. The experiment shows that our I-LLM achieves comparable accuracy to the FP baseline and outperforms non-integer quantization methods. For example, I-LLM can operate at W4A4 with negligible loss of accuracy. To our knowledge, we are the first to bridge the gap between integer-only quantization and LLMs. We've published our code on anonymous.4open.science, aiming to contribute to the advancement of this field.
Authors: Yu Chen, Edoardo Patelli, Zhen Yang, Adolphus Lye
Abstract: Advances in Deep Learning bring further investigation into credibility and robustness, especially for safety-critical engineering applications such as the nuclear industry. The key challenges include the availability of data set (often scarce and sparse) and insufficient consideration of the uncertainty in the data, model, and prediction. This paper therefore presents a meta-learning based approach that is both uncertainty- and prior knowledge-informed, aiming at trustful predictions of material properties for the nuclear reactor design. It is suited for robust learning under limited data. Uncertainty has been accounted for where a distribution of predictor functions are produced for extrapolation. Results suggest it achieves superior performance than existing empirical methods in rupture life prediction, a case which is typically under a small data regime. While demonstrated herein with rupture properties, this learning approach is transferable to solve similar problems of data scarcity across the nuclear industry. It is of great importance to boosting the AI analytics in the nuclear industry by proving the applicability and robustness while providing tools that can be trusted.
Authors: D. van der Sluis
Abstract: To investigate whether "Intelligence is the capacity of an information-processing system to adapt to its environment while operating with insufficient knowledge and resources", we look at utilising the non axiomatic reasoning system (NARS) for speech recognition. This article presents NUTS: raNdom dimensionality redUction non axiomaTic reasoning few Shot learner for perception. NUTS consists of naive dimensionality reduction, some pre-processing, and then non axiomatic reasoning (NARS). With only 2 training examples NUTS performs similarly to the Whisper Tiny model for discrete word identification.
Authors: Yingqi Liu, Yifan Shi, Qinglun Li, Baoyuan Wu, Xueqian Wang, Li Shen
Abstract: Personalized Federated Learning (PFL) is proposed to find the greatest personalized models for each client. To avoid the central failure and communication bottleneck in the server-based FL, we concentrate on the Decentralized Personalized Federated Learning (DPFL) that performs distributed model training in a Peer-to-Peer (P2P) manner. Most personalized works in DPFL are based on undirected and symmetric topologies, however, the data, computation and communication resources heterogeneity result in large variances in the personalized models, which lead the undirected aggregation to suboptimal personalized performance and unguaranteed convergence. To address these issues, we propose a directed collaboration DPFL framework by incorporating stochastic gradient push and partial model personalized, called \textbf{D}ecentralized \textbf{Fed}erated \textbf{P}artial \textbf{G}radient \textbf{P}ush (\textbf{DFedPGP}). It personalizes the linear classifier in the modern deep model to customize the local solution and learns a consensus representation in a fully decentralized manner. Clients only share gradients with a subset of neighbors based on the directed and asymmetric topologies, which guarantees flexible choices for resource efficiency and better convergence. Theoretically, we show that the proposed DFedPGP achieves a superior convergence rate of $\mathcal{O}(\frac{1}{\sqrt{T}})$ in the general non-convex setting, and prove the tighter connectivity among clients will speed up the convergence. The proposed method achieves state-of-the-art (SOTA) accuracy in both data and computation heterogeneity scenarios, demonstrating the efficiency of the directed collaboration and partial gradient push.
Authors: Dongjae Jeon, Wonje Jeung, Taeheon Kim, Albert No, Jonghyun Choi
Abstract: Machine unlearning (MU) addresses privacy concerns by removing information of `forgetting data' samples from trained models. Typically, evaluating MU methods involves comparing unlearned models to those retrained from scratch without forgetting data, using metrics such as membership inference attacks (MIA) and accuracy measurements. These evaluations implicitly assume that if the output logits of the unlearned and retrained models are similar, the unlearned model has successfully forgotten the data. Here, we challenge if this assumption is valid. In particular, we conduct a simple experiment of training only the last layer of a given original model using a novel masked-distillation technique while keeping the rest fixed. Surprisingly, simply altering the last layer yields favorable outcomes in the existing evaluation metrics, while the model does not successfully unlearn the samples or classes. For better evaluating the MU methods, we propose a metric that quantifies the residual information about forgetting data samples in intermediate features using mutual information, called information difference index or IDI for short. The IDI provides a comprehensive evaluation of MU methods by efficiently analyzing the internal structure of DNNs. Our metric is scalable to large datasets and adaptable to various model architectures. Additionally, we present COLapse-and-Align (COLA), a simple contrastive-based method that effectively unlearns intermediate features.
Authors: Lang Feng, Pengjie Gu, Bo An, Gang Pan
Abstract: Diffusion planners have shown promise in handling long-horizon and sparse-reward tasks due to the non-autoregressive plan generation. However, their inherent stochastic risk of generating infeasible trajectories presents significant challenges to their reliability and stability. We introduce a novel approach, the Trajectory Aggregation Tree (TAT), to address this issue in diffusion planners. Compared to prior methods that rely solely on raw trajectory predictions, TAT aggregates information from both historical and current trajectories, forming a dynamic tree-like structure. Each trajectory is conceptualized as a branch and individual states as nodes. As the structure evolves with the integration of new trajectories, unreliable states are marginalized, and the most impactful nodes are prioritized for decision-making. TAT can be deployed without modifying the original training and sampling pipelines of diffusion planners, making it a training-free, ready-to-deploy solution. We provide both theoretical analysis and empirical evidence to support TAT's effectiveness. Our results highlight its remarkable ability to resist the risk from unreliable trajectories, guarantee the performance boosting of diffusion planners in $100\%$ of tasks, and exhibit an appreciable tolerance margin for sample quality, thereby enabling planning with a more than $3\times$ acceleration.
Authors: Byeonghu Na, Yeongmin Kim, Minsang Park, Donghyeok Shin, Wanmo Kang, Il-Chul Moon
Abstract: Recent advances in powerful pre-trained diffusion models encourage the development of methods to improve the sampling performance under well-trained diffusion models. This paper introduces Diffusion Rejection Sampling (DiffRS), which uses a rejection sampling scheme that aligns the sampling transition kernels with the true ones at each timestep. The proposed method can be viewed as a mechanism that evaluates the quality of samples at each intermediate timestep and refines them with varying effort depending on the sample. Theoretical analysis shows that DiffRS can achieve a tighter bound on sampling error compared to pre-trained models. Empirical results demonstrate the state-of-the-art performance of DiffRS on the benchmark datasets and the effectiveness of DiffRS for fast diffusion samplers and large-scale text-to-image diffusion models. Our code is available at https://github.com/aailabkaist/DiffRS.
Authors: Onur Boyar, Yanheng Gu, Yuji Tanaka, Shunsuke Tonogai, Tomoya Itakura, Ichiro Takeuchi
Abstract: Generative modeling of crystal structures is significantly challenged by the complexity of input data, which constrains the ability of these models to explore and discover novel crystals. This complexity often confines de novo design methodologies to merely small perturbations of known crystals and hampers the effective application of advanced optimization techniques. One such optimization technique, Latent Space Bayesian Optimization (LSBO) has demonstrated promising results in uncovering novel objects across various domains, especially when combined with Variational Autoencoders (VAEs). Recognizing LSBO's potential and the critical need for innovative crystal discovery, we introduce Crystal-LSBO, a de novo design framework for crystals specifically tailored to enhance explorability within LSBO frameworks. Crystal-LSBO employs multiple VAEs, each dedicated to a distinct aspect of crystal structure: lattice, coordinates, and chemical elements, orchestrated by an integrative model that synthesizes these components into a cohesive output. This setup not only streamlines the learning process but also produces explorable latent spaces thanks to the decreased complexity of the learning task for each model, enabling LSBO approaches to operate. Our study pioneers the use of LSBO for de novo crystal design, demonstrating its efficacy through optimization tasks focused mainly on formation energy values. Our results highlight the effectiveness of our methodology, offering a new perspective for de novo crystal discovery.
Authors: Yige Hong, Qiaomin Xie, Yudong Chen, Weina Wang
Abstract: We consider the discrete-time infinite-horizon average-reward restless bandit problem. We propose a novel policy that maintains two dynamic subsets of arms: one subset of arms has a nearly optimal state distribution and takes actions according to an Optimal Local Control routine; the other subset of arms is driven towards the optimal state distribution and gradually merged into the first subset. We show that our policy is asymptotically optimal with an $O(\exp(-C N))$ optimality gap for an $N$-armed problem, under the mild assumptions of aperiodic-unichain, non-degeneracy, and local stability. Our policy is the first to achieve exponential asymptotic optimality under the above set of easy-to-verify assumptions, whereas prior work either requires a strong Global Attractor assumption or only achieves an $O(1/\sqrt{N})$ optimality gap. We further discuss the fundamental obstacles in significantly weakening our assumptions. In particular, we prove a lower bound showing that local stability is fundamental for exponential asymptotic optimality.
Authors: Severi Rissanen, Markus Heinonen, Arno Solin
Abstract: In the domains of image and audio, diffusion models have shown impressive performance. However, their application to discrete data types, such as language, has often been suboptimal compared to autoregressive generative models. This paper tackles the challenge of improving discrete diffusion models by introducing a structured forward process that leverages the inherent information hierarchy in discrete categories, such as words in text. Our approach biases the generative process to produce certain categories before others, resulting in a notable improvement in log-likelihood scores on the text8 dataset. This work paves the way for more advances in discrete diffusion models with potentially significant enhancements in performance.
Authors: Donato Crisostomi, Marco Fumero, Daniele Baieri, Florian Bernard, Emanuele Rodol\`a
Abstract: In this paper, we present a novel data-free method for merging neural networks in weight space. Differently from most existing works, our method optimizes for the permutations of network neurons globally across all layers. This allows us to enforce cycle consistency of the permutations when merging $N \geq 3$ models, allowing circular compositions of permutations to be computed without accumulating error along the path. We qualitatively and quantitatively motivate the need for such a constraint, showing its benefits when merging sets of models in scenarios spanning varying architectures and datasets. We finally show that, when coupled with activation renormalization, our approach yields the best results in the task.
Authors: Zhonghang Li, Lianghao Xia, Yong Xu, Chao Huang
Abstract: The objective of traffic prediction is to accurately forecast and analyze the dynamics of transportation patterns, considering both space and time. However, the presence of distribution shift poses a significant challenge in this field, as existing models struggle to generalize well when faced with test data that significantly differs from the training distribution. To tackle this issue, this paper introduces a simple and universal spatio-temporal prompt-tuning framework-FlashST, which adapts pre-trained models to the specific characteristics of diverse downstream datasets, improving generalization in diverse traffic prediction scenarios. Specifically, the FlashST framework employs a lightweight spatio-temporal prompt network for in-context learning, capturing spatio-temporal invariant knowledge and facilitating effective adaptation to diverse scenarios. Additionally, we incorporate a distribution mapping mechanism to align the data distributions of pre-training and downstream data, facilitating effective knowledge transfer in spatio-temporal forecasting. Empirical evaluations demonstrate the effectiveness of our FlashST across different spatio-temporal prediction tasks using diverse urban datasets. Code is available at https://github.com/HKUDS/FlashST.
Authors: Xiumei Deng, Jun Li, Long Shi, Kang Wei, Ming Ding, Yumeng Shao, Wen Chen, Shi Jin
Abstract: Digital twin (DT) has emerged as a promising solution to enhance manufacturing efficiency in industrial Internet of Things (IIoT) networks. To promote the efficiency and trustworthiness of DT for wireless IIoT networks, we propose a blockchain-enabled DT (B-DT) framework that employs deep neural network (DNN) partitioning technique and reputation-based consensus mechanism, wherein the DTs maintained at the gateway side execute DNN inference tasks using the data collected from their associated IIoT devices. First, we employ DNN partitioning technique to offload the top-layer DNN inference tasks to the access point (AP) side, which alleviates the computation burden at the gateway side and thereby improves the efficiency of DNN inference. Second, we propose a reputation-based consensus mechanism that integrates Proof of Work (PoW) and Proof of Stake (PoS). Specifically, the proposed consensus mechanism evaluates the off-chain reputation of each AP according to its computation resource contributions to the DNN inference tasks, and utilizes the off-chain reputation as a stake to adjust the block generation difficulty. Third, we formulate a stochastic optimization problem of communication resource (i.e., partition point) and computation resource allocation (i.e., computation frequency of APs for top-layer DNN inference and block generation) to minimize system latency under the time-varying channel state and long-term constraints of off-chain reputation, and solve the problem using Lyapunov optimization method. Experimental results show that the proposed dynamic DNN partitioning and resource allocation (DPRA) algorithm outperforms the baselines in terms of reducing the overall latency while guaranteeing the trustworthiness of the B-DT system.
Authors: Dong Bok Lee, Aoxuan Silvia Zhang, Byungjoo Kim, Junhyeon Park, Juho Lee, Sung Ju Hwang, Hae Beom Lee
Abstract: In this paper, we address the problem of cost-sensitive multi-fidelity Bayesian Optimization (BO) for efficient hyperparameter optimization (HPO). Specifically, we assume a scenario where users want to early-stop the BO when the performance improvement is not satisfactory with respect to the required computational cost. Motivated by this scenario, we introduce utility, which is a function predefined by each user and describes the trade-off between cost and performance of BO. This utility function, combined with our novel acquisition function and stopping criterion, allows us to dynamically choose for each BO step the best configuration that we expect to maximally improve the utility in future, and also automatically stop the BO around the maximum utility. Further, we improve the sample efficiency of existing learning curve (LC) extrapolation methods with transfer learning, while successfully capturing the correlations between different configurations to develop a sensible surrogate function for multi-fidelity BO. We validate our algorithm on various LC datasets and found it outperform all the previous multi-fidelity BO and transfer-BO baselines we consider, achieving significantly better trade-off between cost and performance of BO.
Authors: Xiumei Deng, Jun Li, Kang Wei, Long Shi, Zeihui Xiong, Ming Ding, Wen Chen, Shi Jin, H. Vincent Poor
Abstract: Adaptive moment estimation (Adam), as a Stochastic Gradient Descent (SGD) variant, has gained widespread popularity in federated learning (FL) due to its fast convergence. However, federated Adam (FedAdam) algorithms suffer from a threefold increase in uplink communication overhead compared to federated SGD (FedSGD) algorithms, which arises from the necessity to transmit both local model updates and first and second moment estimates from distributed devices to the centralized server for aggregation. Driven by this issue, we propose a novel sparse FedAdam algorithm called FedAdam-SSM, wherein distributed devices sparsify the updates of local model parameters and moment estimates and subsequently upload the sparse representations to the centralized server. To further reduce the communication overhead, the updates of local model parameters and moment estimates incorporate a shared sparse mask (SSM) into the sparsification process, eliminating the need for three separate sparse masks. Theoretically, we develop an upper bound on the divergence between the local model trained by FedAdam-SSM and the desired model trained by centralized Adam, which is related to sparsification error and imbalanced data distribution. By minimizing the divergence bound between the model trained by FedAdam-SSM and centralized Adam, we optimize the SSM to mitigate the learning performance degradation caused by sparsification error. Additionally, we provide convergence bounds for FedAdam-SSM in both convex and non-convex objective function settings, and investigate the impact of local epoch, learning rate and sparsification ratio on the convergence rate of FedAdam-SSM. Experimental results show that FedAdam-SSM outperforms baselines in terms of convergence rate (over 1.1$\times$ faster than the sparse FedAdam baselines) and test accuracy (over 14.5\% ahead of the quantized FedAdam baselines).
Authors: Seong-Hyeon Hwang, Minsu Kim, Steven Euijong Whang
Abstract: We study the problem of robust data augmentation for regression tasks in the presence of noisy data. Data augmentation is essential for generalizing deep learning models, but most of the techniques like the popular Mixup are primarily designed for classification tasks on image data. Recently, there are also Mixup techniques that are specialized to regression tasks like C-Mixup. In comparison to Mixup, which takes linear interpolations of pairs of samples, C-Mixup is more selective in which samples to mix based on their label distances for better regression performance. However, C-Mixup does not distinguish noisy versus clean samples, which can be problematic when mixing and lead to suboptimal model performance. At the same time, robust training has been heavily studied where the goal is to train accurate models against noisy data through multiple rounds of model training. We thus propose our data augmentation strategy RC-Mixup, which tightly integrates C-Mixup with multi-round robust training methods for a synergistic effect. In particular, C-Mixup improves robust training in identifying clean data, while robust training provides cleaner data to C-Mixup for it to perform better. A key advantage of RC-Mixup is that it is data-centric where the robust model training algorithm itself does not need to be modified, but can simply benefit from data mixing. We show in our experiments that RC-Mixup significantly outperforms C-Mixup and robust training baselines on noisy data benchmarks and can be integrated with various robust training methods.
Authors: Leon G\"otz, Marcel Kollovieh, Stephan G\"unnemann, Leo Schwinn
Abstract: Transformer architectures have shown promising results in time series processing. However, despite recent advances in subquadratic attention mechanisms or state-space models, processing very long sequences still imposes significant computational requirements. Token merging, which involves replacing multiple tokens with a single one calculated as their linear combination, has shown to considerably improve the throughput of vision transformer architectures while maintaining accuracy. In this work, we go beyond computer vision and perform the first investigations of token merging in time series analysis on both time series transformers and state-space models. To effectively scale token merging to long sequences, we introduce local merging, a domain-specific token merging algorithm that selectively combines tokens within a local neighborhood, adjusting the computational complexity from linear to quadratic based on the neighborhood size. Our comprehensive empirical evaluation demonstrates that token merging offers substantial computational benefits with minimal impact on accuracy across various models and datasets. On the recently proposed Chronos foundation model, we achieve accelerations up to 5400% with only minor accuracy degradations.
Authors: Ruo-Chun Tzeng, Naoto Ohsaka, Kaito Ariu
Abstract: We study the matroid semi-bandits problem, where at each round the learner plays a subset of $K$ arms from a feasible set, and the goal is to maximize the expected cumulative linear rewards. Existing algorithms have per-round time complexity at least $\Omega(K)$, which becomes expensive when $K$ is large. To address this computational issue, we propose FasterCUCB whose sampling rule takes time sublinear in $K$ for common classes of matroids: $O(D\text{ polylog}(K)\text{ polylog}(T))$ for uniform matroids, partition matroids, and graphical matroids, and $O(D\sqrt{K}\text{ polylog}(T))$ for transversal matroids. Here, $D$ is the maximum number of elements in any feasible subset of arms, and $T$ is the horizon. Our technique is based on dynamic maintenance of an approximate maximum-weight basis over inner-product weights. Although the introduction of an approximate maximum-weight basis presents a challenge in regret analysis, we can still guarantee an upper bound on regret as tight as CUCB in the sense that it matches the gap-dependent lower bound by Kveton et al. (2014a) asymptotically.
Authors: Dean Brandner, Sergio Lucia
Abstract: Model predictive control can optimally deal with nonlinear systems under consideration of constraints. The control performance depends on the model accuracy and the prediction horizon. Recent advances propose to use reinforcement learning applied to a parameterized model predictive controller to recover the optimal control performance even if an imperfect model or short prediction horizons are used. However, common reinforcement learning algorithms rely on first order updates, which only have a linear convergence rate and hence need an excessive amount of dynamic data. Higher order updates are typically intractable if the policy is approximated with neural networks due to the large number of parameters. In this work, we use a parameterized model predictive controller as policy, and leverage the small amount of necessary parameters to propose a trust-region constrained Quasi-Newton training algorithm for policy optimization with a superlinear convergence rate. We show that the required second order derivative information can be calculated by the solution of a linear system of equations. A simulation study illustrates that the proposed training algorithm outperforms other algorithms in terms of data efficiency and accuracy.
Authors: Xiaoting Lyu, Yufei Han, Wei Wang, Hangwei Qian, Ivor Tsang, Xiangliang Zhang
Abstract: Graph Prompt Learning (GPL) bridges significant disparities between pretraining and downstream applications to alleviate the knowledge transfer bottleneck in real-world graph learning. While GPL offers superior effectiveness in graph knowledge transfer and computational efficiency, the security risks posed by backdoor poisoning effects embedded in pretrained models remain largely unexplored. Our study provides a comprehensive analysis of GPL's vulnerability to backdoor attacks. We introduce \textit{CrossBA}, the first cross-context backdoor attack against GPL, which manipulates only the pretraining phase without requiring knowledge of downstream applications. Our investigation reveals both theoretically and empirically that tuning trigger graphs, combined with prompt transformations, can seamlessly transfer the backdoor threat from pretrained encoders to downstream applications. Through extensive experiments involving 3 representative GPL methods across 5 distinct cross-context scenarios and 5 benchmark datasets of node and graph classification tasks, we demonstrate that \textit{CrossBA} consistently achieves high attack success rates while preserving the functionality of downstream applications over clean input. We also explore potential countermeasures against \textit{CrossBA} and conclude that current defenses are insufficient to mitigate \textit{CrossBA}. Our study highlights the persistent backdoor threats to GPL systems, raising trustworthiness concerns in the practices of GPL techniques.
Authors: Wanlin Cai, Kun Wang, Hao Wu, Xiaoxu Chen, Yuankai Wu
Abstract: The challenge of effectively learning inter-series correlations for multivariate time series forecasting remains a substantial and unresolved problem. Traditional deep learning models, which are largely dependent on the Transformer paradigm for modeling long sequences, often fail to integrate information from multiple time series into a coherent and universally applicable model. To bridge this gap, our paper presents ForecastGrapher, a framework reconceptualizes multivariate time series forecasting as a node regression task, providing a unique avenue for capturing the intricate temporal dynamics and inter-series correlations. Our approach is underpinned by three pivotal steps: firstly, generating custom node embeddings to reflect the temporal variations within each series; secondly, constructing an adaptive adjacency matrix to encode the inter-series correlations; and thirdly, augmenting the GNNs' expressive power by diversifying the node feature distribution. To enhance this expressive power, we introduce the Group Feature Convolution GNN (GFC-GNN). This model employs a learnable scaler to segment node features into multiple groups and applies one-dimensional convolutions with different kernel lengths to each group prior to the aggregation phase. Consequently, the GFC-GNN method enriches the diversity of node feature distribution in a fully end-to-end fashion. Through extensive experiments and ablation studies, we show that ForecastGrapher surpasses strong baselines and leading published techniques in the domain of multivariate time series forecasting.
Authors: Omar Erak, Omar Alhussein, Shimaa Naser, Nouf Alabbasi, De Mi, Sami Muhaidat
Abstract: This paper proposes a novel framework that leverages large language models (LLMs) to automate curriculum design, thereby enhancing the application of reinforcement learning (RL) in mobile networks. As mobile networks evolve towards the 6G era, managing their increasing complexity and dynamic nature poses significant challenges. Conventional RL approaches often suffer from slow convergence and poor generalization due to conflicting objectives and the large state and action spaces associated with mobile networks. To address these shortcomings, we introduce curriculum learning, a method that systematically exposes the RL agent to progressively challenging tasks, improving convergence and generalization. However, curriculum design typically requires extensive domain knowledge and manual human effort. Our framework mitigates this by utilizing the generative capabilities of LLMs to automate the curriculum design process, significantly reducing human effort while improving the RL agent's convergence and performance. We deploy our approach within a simulated mobile network environment and demonstrate improved RL convergence rates, generalization to unseen scenarios, and overall performance enhancements. As a case study, we consider autonomous coordination and user association in mobile networks. Our obtained results highlight the potential of combining LLM-based curriculum generation with RL for managing next-generation wireless networks, marking a significant step towards fully autonomous network operations.
Authors: Thanh Trung Huynh, Trong Bang Nguyen, Phi Le Nguyen, Thanh Tam Nguyen, Matthias Weidlich, Quoc Viet Hung Nguyen, Karl Aberer
Abstract: Federated learning (FL) has recently emerged as a compelling machine learning paradigm, prioritizing the protection of privacy for training data. The increasing demand to address issues such as ``the right to be forgotten'' and combat data poisoning attacks highlights the importance of techniques, known as \textit{unlearning}, which facilitate the removal of specific training data from trained FL models. Despite numerous unlearning methods proposed for centralized learning, they often prove inapplicable to FL due to fundamental differences in the operation of the two learning paradigms. Consequently, unlearning in FL remains in its early stages, presenting several challenges. Many existing unlearning solutions in FL require a costly retraining process, which can be burdensome for clients. Moreover, these methods are primarily validated through experiments, lacking theoretical assurances. In this study, we introduce Fast-FedUL, a tailored unlearning method for FL, which eliminates the need for retraining entirely. Through meticulous analysis of the target client's influence on the global model in each round, we develop an algorithm to systematically remove the impact of the target client from the trained model. In addition to presenting empirical findings, we offer a theoretical analysis delineating the upper bound of our unlearned model and the exact retrained model (the one obtained through retraining using untargeted clients). Experimental results with backdoor attack scenarios indicate that Fast-FedUL effectively removes almost all traces of the target client, while retaining the knowledge of untargeted clients (obtaining a high accuracy of up to 98\% on the main task). Significantly, Fast-FedUL attains the lowest time complexity, providing a speed that is 1000 times faster than retraining. Our source code is publicly available at \url{https://github.com/thanhtrunghuynh93/fastFedUL}.
Authors: Panagiotis Koromilas, Giorgos Bouritsas, Theodoros Giannakopoulos, Mihalis Nicolaou, Yannis Panagakis
Abstract: What do different contrastive learning (CL) losses actually optimize for? Although multiple CL methods have demonstrated remarkable representation learning capabilities, the differences in their inner workings remain largely opaque. In this work, we analyse several CL families and prove that, under certain conditions, they admit the same minimisers when optimizing either their batch-level objectives or their expectations asymptotically. In both cases, an intimate connection with the hyperspherical energy minimisation (HEM) problem resurfaces. Drawing inspiration from this, we introduce a novel CL objective, coined Decoupled Hyperspherical Energy Loss (DHEL). DHEL simplifies the problem by decoupling the target hyperspherical energy from the alignment of positive examples while preserving the same theoretical guarantees. Going one step further, we show the same results hold for another relevant CL family, namely kernel contrastive learning (KCL), with the additional advantage of the expected loss being independent of batch size, thus identifying the minimisers in the non-asymptotic regime. Empirical results demonstrate improved downstream performance and robustness across combinations of different batch sizes and hyperparameters and reduced dimensionality collapse, on several computer vision datasets.
Authors: Christopher Rae, Joseph K. L. Lee, James Richings
Abstract: As Deep Neural Networks (DNNs) grow in size and complexity, they often exceed the memory capacity of a single accelerator, necessitating the sharding of model parameters across multiple accelerators. Pipeline parallelism is a commonly used sharding strategy for training large DNNs. However, current implementations of pipeline parallelism are being unintentionally bottlenecked by the automatic differentiation tools provided by ML frameworks. This paper introduces 2-stage backpropagation (2BP). By splitting the backward propagation step into two separate stages, we can reduce idle compute time. We tested 2BP on various model architectures and pipelining schedules, achieving increases in throughput in all cases. Using 2BP, we were able to achieve a 1.70x increase in throughput compared to traditional methods when training a LLaMa-like transformer with 7 billion parameters across 4 GPUs.
Authors: Tim Po\v{s}tuvan, Claas Grohnfeldt, Michele Russo, Giulio Lovisotto
Abstract: Anomaly detection in continuous-time dynamic graphs is an emerging field yet under-explored in the context of learning-based approaches. In this paper, we pioneer structured analyses of link-level anomalies and graph representation learning for identifying anomalous links in these graphs. First, we introduce a fine-grain taxonomy for edge-level anomalies leveraging structural, temporal, and contextual graph properties. We present a method for generating and injecting such typed anomalies into graphs. Next, we introduce a novel method to generate continuous-time dynamic graphs with consistent patterns across time, structure, and context. To allow temporal graph methods to learn the link anomaly detection task, we extend the generic link prediction setting by: (1) conditioning link existence on contextual edge attributes; and (2) refining the training regime to accommodate diverse perturbations in the negative edge sampler. Building on this, we benchmark methods for anomaly detection. Comprehensive experiments on synthetic and real-world datasets -- featuring synthetic and labeled organic anomalies and employing six state-of-the-art learning methods -- validate our taxonomy and generation processes for anomalies and benign graphs, as well as our approach to adapting link prediction methods for anomaly detection. Our results further reveal that different learning methods excel in capturing different aspects of graph normality and detecting different types of anomalies. We conclude with a comprehensive list of findings highlighting opportunities for future research.
Authors: Albin Soutif--Cormerais, Simone Magistri, Joost van de Weijer, Andew D. Bagdanov
Abstract: Broad, open source availability of large pretrained foundation models on the internet through platforms such as HuggingFace has taken the world of practical deep learning by storm. A classical pipeline for neural network training now typically consists of finetuning these pretrained network on a small target dataset instead of training from scratch. In the case of large models this can be done even on modest hardware using a low rank training technique known as Low-Rank Adaptation (LoRA). While Low Rank training has already been studied in the continual learning setting, existing works often consider storing the learned adapter along with the existing model but rarely attempt to modify the weights of the pretrained model by merging the LoRA with the existing weights after finishing the training of each task. In this article we investigate this setting and study the impact of LoRA rank on the forgetting of the pretraining foundation task and on the plasticity and forgetting of subsequent ones. We observe that this rank has an important impact on forgetting of both the pretraining and downstream tasks. We also observe that vision transformers finetuned in that way exhibit a sort of ``contextual'' forgetting, a behaviour that we do not observe for residual networks and that we believe has not been observed yet in previous continual learning works.
Authors: Nata\v{s}a Tagasovska, Vladimir Gligorijevi\'c, Kyunghyun Cho, Andreas Loukas
Abstract: Across scientific domains, generating new models or optimizing existing ones while meeting specific criteria is crucial. Traditional machine learning frameworks for guided design use a generative model and a surrogate model (discriminator), requiring large datasets. However, real-world scientific applications often have limited data and complex landscapes, making data-hungry models inefficient or impractical. We propose a new framework, PropEn, inspired by ``matching'', which enables implicit guidance without training a discriminator. By matching each sample with a similar one that has a better property value, we create a larger training dataset that inherently indicates the direction of improvement. Matching, combined with an encoder-decoder architecture, forms a domain-agnostic generative framework for property enhancement. We show that training with a matched dataset approximates the gradient of the property of interest while remaining within the data distribution, allowing efficient design optimization. Extensive evaluations in toy problems and scientific applications, such as therapeutic protein design and airfoil optimization, demonstrate PropEn's advantages over common baselines. Notably, the protein design results are validated with wet lab experiments, confirming the competitiveness and effectiveness of our approach.
Authors: Daniel Vranje\v{s}, Oliver Niggemann
Abstract: Empirical research plays a fundamental role in the machine learning domain. At the heart of impactful empirical research lies the development of clear research hypotheses, which then shape the design of experiments. The execution of experiments must be carried out with precision to ensure reliable results, followed by statistical analysis to interpret these outcomes. This process is key to either supporting or refuting initial hypotheses. Despite its importance, there is a high variability in research practices across the machine learning community and no uniform understanding of quality criteria for empirical research. To address this gap, we propose a model for the empirical research process, accompanied by guidelines to uphold the validity of empirical research. By embracing these recommendations, greater consistency, enhanced reliability and increased impact can be achieved.
Authors: Shengchao Hu, Ziqing Fan, Li Shen, Ya Zhang, Yanfeng Wang, Dacheng Tao
Abstract: The purpose of offline multi-task reinforcement learning (MTRL) is to develop a unified policy applicable to diverse tasks without the need for online environmental interaction. Recent advancements approach this through sequence modeling, leveraging the Transformer architecture's scalability and the benefits of parameter sharing to exploit task similarities. However, variations in task content and complexity pose significant challenges in policy formulation, necessitating judicious parameter sharing and management of conflicting gradients for optimal policy performance. In this work, we introduce the Harmony Multi-Task Decision Transformer (HarmoDT), a novel solution designed to identify an optimal harmony subspace of parameters for each task. We approach this as a bi-level optimization problem, employing a meta-learning framework that leverages gradient-based techniques. The upper level of this framework is dedicated to learning a task-specific mask that delineates the harmony subspace, while the inner level focuses on updating parameters to enhance the overall performance of the unified policy. Empirical evaluations on a series of benchmarks demonstrate the superiority of HarmoDT, verifying the effectiveness of our approach.
Authors: Sebastien Origer, Dario Izzo
Abstract: Inspired by the versatility of sinusoidal representation networks (SIRENs), we present a modified Guidance & Control Networks (G&CNETs) variant using periodic activation functions in the hidden layers. We demonstrate that the resulting G&CNETs train faster and achieve a lower overall training error on three different control scenarios on which G&CNETs have been tested previously. A preliminary analysis is presented in an attempt to explain the superior performance of the SIREN architecture for the particular types of tasks that G&CNETs excel on.
Authors: Onno Eberhard, Claire Vernade, Michael Muehlebach
Abstract: Reinforcement learning has traditionally focused on learning state-dependent policies to solve optimal control problems in a closed-loop fashion. In this work, we introduce the paradigm of open-loop reinforcement learning where a fixed action sequence is learned instead. We present three new algorithms: one robust model-based method and two sample-efficient model-free methods. Rather than basing our algorithms on Bellman's equation from dynamic programming, our work builds on Pontryagin's principle from the theory of open-loop optimal control. We provide convergence guarantees and evaluate all methods empirically on a pendulum swing-up task, as well as on two high-dimensional MuJoCo tasks, demonstrating remarkable performance compared to existing baselines.
Authors: Xinran Li, Zifan Liu, Shibo Chen, Jun Zhang
Abstract: In multi-agent reinforcement learning (MARL), effective exploration is critical, especially in sparse reward environments. Although introducing global intrinsic rewards can foster exploration in such settings, it often complicates credit assignment among agents. To address this difficulty, we propose Individual Contributions as intrinsic Exploration Scaffolds (ICES), a novel approach to motivate exploration by assessing each agent's contribution from a global view. In particular, ICES constructs exploration scaffolds with Bayesian surprise, leveraging global transition information during centralized training. These scaffolds, used only in training, help to guide individual agents towards actions that significantly impact the global latent state transitions. Additionally, ICES separates exploration policies from exploitation policies, enabling the former to utilize privileged global information during training. Extensive experiments on cooperative benchmark tasks with sparse rewards, including Google Research Football (GRF) and StarCraft Multi-agent Challenge (SMAC), demonstrate that ICES exhibits superior exploration capabilities compared with baselines. The code is publicly available at https://github.com/LXXXXR/ICES.
Authors: Antonin Joly, Nicolas Keriven
Abstract: Graph coarsening aims to reduce the size of a large graph while preserving some of its key properties, which has been used in many applications to reduce computational load and memory footprint. For instance, in graph machine learning, training Graph Neural Networks (GNNs) on coarsened graphs leads to drastic savings in time and memory. However, GNNs rely on the Message-Passing (MP) paradigm, and classical spectral preservation guarantees for graph coarsening do not directly lead to theoretical guarantees when performing naive message-passing on the coarsened graph. In this work, we propose a new message-passing operation specific to coarsened graphs, which exhibit theoretical guarantees on the preservation of the propagated signal. Interestingly, and in a sharp departure from previous proposals, this operation on coarsened graphs is oriented, even when the original graph is undirected. We conduct node classification tasks on synthetic and real data and observe improved results compared to performing naive message-passing on the coarsened graph.
Authors: Kazuki Egashira, Mark Vero, Robin Staab, Jingxuan He, Martin Vechev
Abstract: Quantization leverages lower-precision weights to reduce the memory usage of large language models (LLMs) and is a key technique for enabling their deployment on commodity hardware. While LLM quantization's impact on utility has been extensively explored, this work for the first time studies its adverse effects from a security perspective. We reveal that widely used quantization methods can be exploited to produce a harmful quantized LLM, even though the full-precision counterpart appears benign, potentially tricking users into deploying the malicious quantized model. We demonstrate this threat using a three-staged attack framework: (i) first, we obtain a malicious LLM through fine-tuning on an adversarial task; (ii) next, we quantize the malicious model and calculate constraints that characterize all full-precision models that map to the same quantized model; (iii) finally, using projected gradient descent, we tune out the poisoned behavior from the full-precision model while ensuring that its weights satisfy the constraints computed in step (ii). This procedure results in an LLM that exhibits benign behavior in full precision but when quantized, it follows the adversarial behavior injected in step (i). We experimentally demonstrate the feasibility and severity of such an attack across three diverse scenarios: vulnerable code generation, content injection, and over-refusal attack. In practice, the adversary could host the resulting full-precision model on an LLM community hub such as Hugging Face, exposing millions of users to the threat of deploying its malicious quantized version on their devices.
Authors: Sike Wang, Jia Li, Pan Zhou, Hua Huang
Abstract: Second-order optimizers, maintaining a matrix termed a preconditioner, are superior to first-order optimizers in both theory and practice. The states forming the preconditioner and its inverse root restrict the maximum size of models trained by second-order optimizers. To address this, compressing 32-bit optimizer states to lower bitwidths has shown promise in reducing memory usage. However, current approaches only pertain to first-order optimizers. In this paper, we propose the first 4-bit second-order optimizers, exemplified by 4-bit Shampoo, maintaining performance similar to that of 32-bit ones. We show that quantizing the eigenvector matrix of the preconditioner in 4-bit Shampoo is remarkably better than quantizing the preconditioner itself both theoretically and experimentally. By rectifying the orthogonality of the quantized eigenvector matrix, we enhance the approximation of the preconditioner's eigenvector matrix, which also benefits the computation of its inverse 4-th root. Besides, we find that linear square quantization slightly outperforms dynamic tree quantization when quantizing second-order optimizer states. Evaluation on various networks for image classification demonstrates that our 4-bit Shampoo achieves comparable test accuracy to its 32-bit counterpart while being more memory-efficient. The source code will be made available.
Authors: Ang\'eline Pouget, Nikola Jovanovi\'c, Mark Vero, Robin Staab, Martin Vechev
Abstract: The goal of Fair Representation Learning (FRL) is to mitigate biases in machine learning models by learning data representations that enable high accuracy on downstream tasks while minimizing discrimination based on sensitive attributes. The evaluation of FRL methods in many recent works primarily focuses on the tradeoff between downstream fairness and accuracy with respect to a single task that was used to approximate the utility of representations during training (proxy task). This incentivizes retaining only features relevant to the proxy task while discarding all other information. In extreme cases, this can cause the learned representations to collapse to a trivial, binary value, rendering them unusable in transfer settings. In this work, we argue that this approach is fundamentally mismatched with the original motivation of FRL, which arises from settings with many downstream tasks unknown at training time (transfer tasks). To remedy this, we propose to refocus the evaluation protocol of FRL methods primarily around the performance on transfer tasks. A key challenge when conducting such an evaluation is the lack of adequate benchmarks. We address this by formulating four criteria that a suitable evaluation procedure should fulfill. Based on these, we propose TransFair, a benchmark that satisfies these criteria, consisting of novel variations of popular FRL datasets with carefully calibrated transfer tasks. In this setting, we reevaluate state-of-the-art FRL methods, observing that they often overfit to the proxy task, which causes them to underperform on certain transfer tasks. We further highlight the importance of task-agnostic learning signals for FRL methods, as they can lead to more transferrable representations.
Authors: Robert Leppich, Vanessa Borst, Veronika Lesch, Samuel Kounev
Abstract: Time series analysis remains a major challenge due to its sparse characteristics, high dimensionality, and inconsistent data quality. Recent advancements in transformer-based techniques have enhanced capabilities in forecasting and imputation; however, these methods are still resource-heavy, lack adaptability, and face difficulties in integrating both local and global attributes of time series. To tackle these challenges, we propose a new architectural concept for time series analysis based on introspection. Central to this concept is the self-supervised pretraining of Time Series Representation Models (TSRMs), which once learned can be easily tailored and fine-tuned for specific tasks, such as forecasting and imputation, in an automated and resource-efficient manner. Our architecture is equipped with a flexible and hierarchical representation learning process, which is robust against missing data and outliers. It can capture and learn both local and global features of the structure, semantics, and crucial patterns of a given time series category, such as heart rate data. Our learned time series representation models can be efficiently adapted to a specific task, such as forecasting or imputation, without manual intervention. Furthermore, our architecture's design supports explainability by highlighting the significance of each input value for the task at hand. Our empirical study using four benchmark datasets shows that, compared to investigated state-of-the-art baseline methods, our architecture improves imputation and forecasting errors by up to 90.34% and 71.54%, respectively, while reducing the required trainable parameters by up to 92.43%. The source code is available at https://github.com/RobertLeppich/TSRM.
Authors: Longxiang He, Li Shen, Junbo Tan, Xueqian Wang
Abstract: Implicit Q-learning (IQL) serves as a strong baseline for offline RL, which learns the value function using only dataset actions through quantile regression. However, it is unclear how to recover the implicit policy from the learned implicit Q-function and why IQL can utilize weighted regression for policy extraction. IDQL reinterprets IQL as an actor-critic method and gets weights of implicit policy, however, this weight only holds for the optimal value function. In this work, we introduce a different way to solve the implicit policy-finding problem (IPF) by formulating this problem as an optimization problem. Based on this optimization problem, we further propose two practical algorithms AlignIQL and AlignIQL-hard, which inherit the advantages of decoupling actor from critic in IQL and provide insights into why IQL can use weighted regression for policy extraction. Compared with IQL and IDQL, we find our method keeps the simplicity of IQL and solves the implicit policy-finding problem. Experimental results on D4RL datasets show that our method achieves competitive or superior results compared with other SOTA offline RL methods. Especially in complex sparse reward tasks like Antmaze and Adroit, our method outperforms IQL and IDQL by a significant margin.
Authors: Johann Bauer, Sheldon West, Eduardo Alonso, Mark Broom
Abstract: We present two variants of a multi-agent reinforcement learning algorithm based on evolutionary game theoretic considerations. The intentional simplicity of one variant enables us to prove results on its relationship to a system of ordinary differential equations of replicator-mutator dynamics type, allowing us to present proofs on the algorithm's convergence conditions in various settings via its ODE counterpart. The more complicated variant enables comparisons to Q-learning based algorithms. We compare both variants experimentally to WoLF-PHC and frequency-adjusted Q-learning on a range of settings, illustrating cases of increasing dimensionality where our variants preserve convergence in contrast to more complicated algorithms. The availability of analytic results provides a degree of transferability of results as compared to purely empirical case studies, illustrating the general utility of a dynamical systems perspective on multi-agent reinforcement learning when addressing questions of convergence and reliable generalisation.
Authors: Sharut Gupta, Chenyu Wang, Yifei Wang, Tommi Jaakkola, Stefanie Jegelka
Abstract: At the core of self-supervised learning for vision is the idea of learning invariant or equivariant representations with respect to a set of data transformations. This approach, however, introduces strong inductive biases, which can render the representations fragile in downstream tasks that do not conform to these symmetries. In this work, drawing insights from world models, we propose to instead learn a general representation that can adapt to be invariant or equivariant to different transformations by paying attention to context -- a memory module that tracks task-specific states, actions, and future states. Here, the action is the transformation, while the current and future states respectively represent the input's representation before and after the transformation. Our proposed algorithm, Contextual Self-Supervised Learning (ContextSSL), learns equivariance to all transformations (as opposed to invariance). In this way, the model can learn to encode all relevant features as general representations while having the versatility to tail down to task-wise symmetries when given a few examples as the context. Empirically, we demonstrate significant performance gains over existing methods on equivariance-related tasks, supported by both qualitative and quantitative evaluations.
Authors: Youlong Ding, Xueyang Wu, Yining Meng, Yonggang Luo, Hao Wang, Weike Pan
Abstract: Deep learning with differential privacy (DP) has garnered significant attention over the past years, leading to the development of numerous methods aimed at enhancing model accuracy and training efficiency. This paper delves into the problem of training Transformer models with differential privacy. Our treatment is modular: the logic is to `reduce' the problem of training DP Transformer to the more basic problem of training DP vanilla neural nets. The latter is better understood and amenable to many model-agnostic methods. Such `reduction' is done by first identifying the hardness unique to DP Transformer training: the attention distraction phenomenon and a lack of compatibility with existing techniques for efficient gradient clipping. To deal with these two issues, we propose the Re-Attention Mechanism and Phantom Clipping, respectively. We believe that our work not only casts new light on training DP Transformers but also promotes a modular treatment to advance research in the field of differentially private deep learning.
Authors: Kwangjun Ahn, Ashok Cutkosky
Abstract: In this work, we offer a theoretical analysis of two modern optimization techniques for training large and complex models: (i) adaptive optimization algorithms, such as Adam, and (ii) the model exponential moving average (EMA). Specifically, we demonstrate that a clipped version of Adam with model EMA achieves the optimal convergence rates in various nonconvex optimization settings, both smooth and nonsmooth. Moreover, when the scale varies significantly across different coordinates, we demonstrate that the coordinate-wise adaptivity of Adam is provably advantageous. Notably, unlike previous analyses of Adam, our analysis crucially relies on its core elements -- momentum and discounting factors -- as well as model EMA, motivating their wide applications in practice.
Authors: Ismail Nejjar, Faez Ahmed, Olga Fink
Abstract: Regression models often fail to generalize effectively in regions characterized by highly imbalanced label distributions. Previous methods for deep imbalanced regression rely on gradient-based weight updates, which tend to overfit in underrepresented regions. This paper proposes a paradigm shift towards in-context learning as an effective alternative to conventional in-weight learning methods, particularly for addressing imbalanced regression. In-context learning refers to the ability of a model to condition itself, given a prompt sequence composed of in-context samples (input-label pairs) alongside a new query input to generate predictions, without requiring any parameter updates. In this paper, we study the impact of the prompt sequence on the model performance from both theoretical and empirical perspectives. We emphasize the importance of localized context in reducing bias within regions of high imbalance. Empirical evaluations across a variety of real-world datasets demonstrate that in-context learning substantially outperforms existing in-weight learning methods in scenarios with high levels of imbalance.
Authors: Christoph Kern, Michael Kim, Angela Zhou
Abstract: Estimating heterogeneous treatment effects is important to tailor treatments to those individuals who would most likely benefit. However, conditional average treatment effect predictors may often be trained on one population but possibly deployed on different, possibly unknown populations. We use methodology for learning multi-accurate predictors to post-process CATE T-learners (differenced regressions) to become robust to unknown covariate shifts at the time of deployment. The method works in general for pseudo-outcome regression, such as the DR-learner. We show how this approach can combine (large) confounded observational and (smaller) randomized datasets by learning a confounded predictor from the observational dataset, and auditing for multi-accuracy on the randomized controlled trial. We show improvements in bias and mean squared error in simulations with increasingly larger covariate shift, and on a semi-synthetic case study of a parallel large observational study and smaller randomized controlled experiment. Overall, we establish a connection between methods developed for multi-distribution learning and achieve appealing desiderata (e.g. external validity) in causal inference and machine learning.
Authors: Naveen Raman, Mateo Espinosa Zarlenga, Mateja Jamnik
Abstract: Concept-based explainability methods provide insight into deep learning systems by constructing explanations using human-understandable concepts. While the literature on human reasoning demonstrates that we exploit relationships between concepts when solving tasks, it is unclear whether concept-based methods incorporate the rich structure of inter-concept relationships. We analyse the concept representations learnt by concept-based models to understand whether these models correctly capture inter-concept relationships. First, we empirically demonstrate that state-of-the-art concept-based models produce representations that lack stability and robustness, and such methods fail to capture inter-concept relationships. Then, we develop a novel algorithm which leverages inter-concept relationships to improve concept intervention accuracy, demonstrating how correctly capturing inter-concept relationships can improve downstream tasks.
Authors: Yang Zhang, Yawei Li, Xinpeng Wang, Qianli Shen, Barbara Plank, Bernd Bischl, Mina Rezaei, Kenji Kawaguchi
Abstract: Overparametrized transformer networks are the state-of-the-art architecture for Large Language Models (LLMs). However, such models contain billions of parameters making large compute a necessity, while raising environmental concerns. To address these issues, we propose FinerCut, a new form of fine-grained layer pruning, which in contrast to prior work at the transformer block level, considers all self-attention and feed-forward network (FFN) layers within blocks as individual pruning candidates. FinerCut prunes layers whose removal causes minimal alternation to the model's output -- contributing to a new, lean, interpretable, and task-agnostic pruning method. Tested across 9 benchmarks, our approach retains 90% performance of Llama3-8B with 25% layers removed, and 95% performance of Llama3-70B with 30% layers removed, all without fine-tuning or post-pruning reconstruction. Strikingly, we observe intriguing results with FinerCut: 42% (34 out of 80) of the self-attention layers in Llama3-70B can be removed while preserving 99% of its performance -- without additional fine-tuning after removal. Moreover, FinerCut provides a tool to inspect the types and locations of pruned layers, allowing to observe interesting pruning behaviors. For instance, we observe a preference for pruning self-attention layers, often at deeper consecutive decoder layers. We hope our insights inspire future efficient LLM architecture designs.
Authors: Camille Castera, Peter Ochs
Abstract: Towards designing learned optimization algorithms that are usable beyond their training setting, we identify key principles that classical algorithms obey, but have up to now, not been used for Learning to Optimize (L2O). Following these principles, we provide a general design pipeline, taking into account data, architecture and learning strategy, and thereby enabling a synergy between classical optimization and L2O, resulting in a philosophy of Learning Optimization Algorithms. As a consequence our learned algorithms perform well far beyond problems from the training distribution. We demonstrate the success of these novel principles by designing a new learning-enhanced BFGS algorithm and provide numerical experiments evidencing its adaptation to many settings at test time.
Authors: Zhankun Luo, Abolfazl Hashemi
Abstract: We study the trajectory of iterations and the convergence rates of the Expectation-Maximization (EM) algorithm for two-component Mixed Linear Regression (2MLR). The fundamental goal of MLR is to learn the regression models from unlabeled observations. The EM algorithm finds extensive applications in solving the mixture of linear regressions. Recent results have established the super-linear convergence of EM for 2MLR in the noiseless and high SNR settings under some assumptions and its global convergence rate with random initialization has been affirmed. However, the exponent of convergence has not been theoretically estimated and the geometric properties of the trajectory of EM iterations are not well-understood. In this paper, first, using Bessel functions we provide explicit closed-form expressions for the EM updates under all SNR regimes. Then, in the noiseless setting, we completely characterize the behavior of EM iterations by deriving a recurrence relation at the population level and notably show that all the iterations lie on a certain cycloid. Based on this new trajectory-based analysis, we exhibit the theoretical estimate for the exponent of super-linear convergence and further improve the statistical error bound at the finite-sample level. Our analysis provides a new framework for studying the behavior of EM for Mixed Linear Regression.
Authors: Shuran Zheng, Yongchan Kwon, Xuan Qi, James Zou
Abstract: A common way to evaluate a dataset in ML involves training a model on this dataset and assessing the model's performance on a test set. However, this approach has two issues: (1) it may incentivize undesirable data manipulation in data marketplaces, as the self-interested data providers seek to modify the dataset to maximize their evaluation scores; (2) it may select datasets that overfit to potentially small test sets. We propose a new data valuation method that provably guarantees the following: data providers always maximize their expected score by truthfully reporting their observed data. Any manipulation of the data, including but not limited to data duplication, adding random data, data removal, or re-weighting data from different groups, cannot increase their expected score. Our method, following the paradigm of proper scoring rules, measures the pointwise mutual information (PMI) of the test dataset and the evaluated dataset. However, computing the PMI of two datasets is challenging. We introduce a novel PMI measuring method that greatly improves tractability within Bayesian machine learning contexts. This is accomplished through a new characterization of PMI that relies solely on the posterior probabilities of the model parameter at an arbitrarily selected value. Finally, we support our theoretical results with simulations and further test the effectiveness of our data valuation method in identifying the top datasets among multiple data providers. Interestingly, our method outperforms the standard approach of selecting datasets based on the trained model's test performance, suggesting that our truthful valuation score can also be more robust to overfitting.
Authors: Antonios Valkanas, Boris N. Oreshkin, Mark Coates
Abstract: Online deep learning solves the problem of learning from streams of data, reconciling two opposing objectives: learn fast and learn deep. Existing work focuses almost exclusively on exploring pure deep learning solutions, which are much better suited to handle the "deep" than the "fast" part of the online learning equation. In our work, we propose a different paradigm, based on a hybrid multilearner approach. First, we develop a fast online logistic regression learner. This learner does not rely on backpropagation. Instead, it uses closed form recursive updates of model parameters, handling the fast learning part of the online learning problem. We then analyze the existing online deep learning theory and show that the widespread ODL approach, currently operating at complexity $O(L^2)$ in terms of the number of layers $L$, can be equivalently implemented in $O(L)$ complexity. This further leads us to the cascaded multilearner design, in which multiple shallow and deep learners are co-trained to solve the online learning problem in a cooperative, synergistic fashion. We show that this approach achieves state-of-the-art results on common online learning datasets, while also being able to handle missing features gracefully. Our code is publicly available at https://github.com/AntonValk/MODL.
Authors: Yuhui Wang, Miroslav Strupl, Francesco Faccio, Qingyuan Wu, Haozhe Liu, Micha{\l} Grudzie\'n, Xiaoyang Tan, J\"urgen Schmidhuber
Abstract: Learning from multi-step off-policy data collected by a set of policies is a core problem of reinforcement learning (RL). Approaches based on importance sampling (IS) often suffer from large variances due to products of IS ratios. Typical IS-free methods, such as $n$-step Q-learning, look ahead for $n$ time steps along the trajectory of actions (where $n$ is called the lookahead depth) and utilize off-policy data directly without any additional adjustment. They work well for proper choices of $n$. We show, however, that such IS-free methods underestimate the optimal value function (VF), especially for large $n$, restricting their capacity to efficiently utilize information from distant future time steps. To overcome this problem, we introduce a novel, IS-free, multi-step off-policy method that avoids the underestimation issue and converges to the optimal VF. At its core lies a simple but non-trivial \emph{highway gate}, which controls the information flow from the distant future by comparing it to a threshold. The highway gate guarantees convergence to the optimal VF for arbitrary $n$ and arbitrary behavioral policies. It gives rise to a novel family of off-policy RL algorithms that safely learn even when $n$ is very large, facilitating rapid credit assignment from the far future to the past. On tasks with greatly delayed rewards, including video games where the reward is given only at the end of the game, our new methods outperform many existing multi-step off-policy algorithms.
Authors: Zihui Wang, Zheng Wang, Lingjuan Lyu, Zhaopeng Peng, Zhicheng Yang, Chenglu Wen, Rongshan Yu, Cheng Wang, Xiaoliang Fan
Abstract: Collaborative fairness stands as an essential element in federated learning to encourage client participation by equitably distributing rewards based on individual contributions. Existing methods primarily focus on adjusting gradient allocations among clients to achieve collaborative fairness. However, they frequently overlook crucial factors such as maintaining consistency across local models and catering to the diverse requirements of high-contributing clients. This oversight inevitably decreases both fairness and model accuracy in practice. To address these issues, we propose FedSAC, a novel Federated learning framework with dynamic Submodel Allocation for Collaborative fairness, backed by a theoretical convergence guarantee. First, we present the concept of "bounded collaborative fairness (BCF)", which ensures fairness by tailoring rewards to individual clients based on their contributions. Second, to implement the BCF, we design a submodel allocation module with a theoretical guarantee of fairness. This module incentivizes high-contributing clients with high-performance submodels containing a diverse range of crucial neurons, thereby preserving consistency across local models. Third, we further develop a dynamic aggregation module to adaptively aggregate submodels, ensuring the equitable treatment of low-frequency neurons and consequently enhancing overall model accuracy. Extensive experiments conducted on three public benchmarks demonstrate that FedSAC outperforms all baseline methods in both fairness and model accuracy. We see this work as a significant step towards incentivizing broader client participation in federated learning. The source code is available at https://github.com/wangzihuixmu/FedSAC.
Authors: Germain Vivier--Ardisson, Alexandre Forel, Axel Parmentier, Thibaut Vidal
Abstract: Optimization layers in deep neural networks have enjoyed a growing popularity in structured learning, improving the state of the art on a variety of applications. Yet, these pipelines lack interpretability since they are made of two opaque layers: a highly non-linear prediction model, such as a deep neural network, and an optimization layer, which is typically a complex black-box solver. Our goal is to improve the transparency of such methods by providing counterfactual explanations. We build upon variational autoencoders a principled way of obtaining counterfactuals: working in the latent space leads to a natural notion of plausibility of explanations. We finally introduce a variant of the classic loss for VAE training that improves their performance in our specific structured context. These provide the foundations of CF-OPT, a first-order optimization algorithm that can find counterfactual explanations for a broad class of structured learning architectures. Our numerical results show that both close and plausible explanations can be obtained for problems from the recent literature.
Authors: Anchit Jain, Rozhin Nobahari, Aristide Baratin, Stefano Sarao Mannelli
Abstract: Machine learning systems often acquire biases by leveraging undesired features in the data, impacting accuracy variably across different sub-populations. Current understanding of bias formation mostly focuses on the initial and final stages of learning, leaving a gap in knowledge regarding the transient dynamics. To address this gap, this paper explores the evolution of bias in a teacher-student setup modeling different data sub-populations with a Gaussian-mixture model. We provide an analytical description of the stochastic gradient descent dynamics of a linear classifier in this setting, which we prove to be exact in high dimension. Notably, our analysis reveals how different properties of sub-populations influence bias at different timescales, showing a shifting preference of the classifier during training. Applying our findings to fairness and robustness, we delineate how and when heterogeneous data and spurious features can generate and amplify bias. We empirically validate our results in more complex scenarios by training deeper networks on synthetic and real datasets, including CIFAR10, MNIST, and CelebA.
Authors: David Anton, Jendrik-Alexander Tr\"oger, Henning Wessels, Ulrich R\"omer, Alexander Henkes, Stefan Hartmann
Abstract: The calibration of constitutive models from full-field data has recently gained increasing interest due to improvements in full-field measurement capabilities. In addition to the experimental characterization of novel materials, continuous structural health monitoring is another application that is of great interest. However, monitoring is usually associated with severe time constraints, difficult to meet with standard numerical approaches. Therefore, parametric physics-informed neural networks (PINNs) for constitutive model calibration from full-field displacement data are investigated. In an offline stage, a parametric PINN can be trained to learn a parameterized solution of the underlying partial differential equation. In the subsequent online stage, the parametric PINN then acts as a surrogate for the parameters-to-state map in calibration. We test the proposed approach for the deterministic least-squares calibration of a linear elastic as well as a hyperelastic constitutive model from noisy synthetic displacement data. We further carry out Markov chain Monte Carlo-based Bayesian inference to quantify the uncertainty. A proper statistical evaluation of the results underlines the high accuracy of the deterministic calibration and that the estimated uncertainty is valid. Finally, we consider experimental data and show that the results are in good agreement with a Finite Element Method-based calibration. Due to the fast evaluation of PINNs, calibration can be performed in near real-time. This advantage is particularly evident in many-query applications such as Markov chain Monte Carlo-based Bayesian inference.
Authors: Mathieu Chevalley, Patrick Schwab, Arash Mehrjou
Abstract: Targeted and uniform interventions to a system are crucial for unveiling causal relationships. While several methods have been developed to leverage interventional data for causal structure learning, their practical application in real-world scenarios often remains challenging. Recent benchmark studies have highlighted these difficulties, even when large numbers of single-variable intervention samples are available. In this work, we demonstrate, both theoretically and empirically, that such datasets contain a wealth of causal information that can be effectively extracted under realistic assumptions about the data distribution. More specifically, we introduce the notion of interventional faithfulness, which relies on comparisons between the marginal distributions of each variable across observational and interventional settings, and we introduce a score on causal orders. Under this assumption, we are able to prove strong theoretical guarantees on the optimum of our score that also hold for large-scale settings. To empirically verify our theory, we introduce Intersort, an algorithm designed to infer the causal order from datasets containing large numbers of single-variable interventions by approximately optimizing our score. Intersort outperforms baselines (GIES, PC and EASE) on almost all simulated data settings replicating common benchmarks in the field. Our proposed novel approach to modeling interventional datasets thus offers a promising avenue for advancing causal inference, highlighting significant potential for further enhancements under realistic assumptions.
Authors: Jihao Andreas Lin, Shreyas Padhy, Bruno Mlodozeniec, Jos\'e Miguel Hern\'andez-Lobato
Abstract: Gaussian processes are a versatile probabilistic machine learning model whose effectiveness often depends on good hyperparameters, which are typically learned by maximising the marginal likelihood. In this work, we consider iterative methods, which use iterative linear system solvers to approximate marginal likelihood gradients up to a specified numerical precision, allowing a trade-off between compute time and accuracy of a solution. We introduce a three-level hierarchy of marginal likelihood optimisation for iterative Gaussian processes, and identify that the computational costs are dominated by solving sequential batches of large positive-definite systems of linear equations. We then propose to amortise computations by reusing solutions of linear system solvers as initialisations in the next step, providing a $\textit{warm start}$. Finally, we discuss the necessary conditions and quantify the consequences of warm starts and demonstrate their effectiveness on regression tasks, where warm starts achieve the same results as the conventional procedure while providing up to a $16 \times$ average speed-up among datasets.
Authors: Ziheng Qin, Zhaopan Xu, Yukun Zhou, Zangwei Zheng, Zebang Cheng, Hao Tang, Lei Shang, Baigui Sun, Xiaojiang Peng, Radu Timofte, Hongxun Yao, Kai Wang, Yang You
Abstract: Deep learning benefits from the growing abundance of available data. Meanwhile, efficiently dealing with the growing data scale has become a challenge. Data publicly available are from different sources with various qualities, and it is impractical to do manual cleaning against noise and redundancy given today's data scale. There are existing techniques for cleaning/selecting the collected data. However, these methods are mainly proposed for offline settings that target one of the cleanness and redundancy problems. In practice, data are growing exponentially with both problems. This leads to repeated data curation with sub-optimal efficiency. To tackle this challenge, we propose InfoGrowth, an efficient online algorithm for data cleaning and selection, resulting in a growing dataset that keeps up to date with awareness of cleanliness and diversity. InfoGrowth can improve data quality/efficiency on both single-modal and multi-modal tasks, with an efficient and scalable design. Its framework makes it practical for real-world data engines.
Authors: Devina Mohan, Anna M. M. Scaife
Abstract: The radio astronomy community is rapidly adopting deep learning techniques to deal with the huge data volumes expected from the next generation of radio observatories. Bayesian neural networks (BNNs) provide a principled way to model uncertainty in the predictions made by such deep learning models and will play an important role in extracting well-calibrated uncertainty estimates on their outputs. In this work, we evaluate the performance of different BNNs against the following criteria: predictive performance, uncertainty calibration and distribution-shift detection for the radio galaxy classification problem.
Authors: Gefan Yang, Elizabeth Louise Baker, Michael L. Severinsen, Christy Anna Hipsley, Stefan Sommer
Abstract: The diffusion bridge is a type of diffusion process that conditions on hitting a specific state within a finite time period. It has broad applications in fields such as Bayesian inference, financial mathematics, control theory, and shape analysis. However, simulating the diffusion bridge for natural data can be challenging due to both the intractability of the drift term and continuous representations of the data. Although several methods are available to simulate finite-dimensional diffusion bridges, infinite-dimensional cases remain unresolved. In the paper, we present a solution to this problem by merging score-matching techniques with operator learning, enabling a direct approach to score-matching for the infinite-dimensional bridge. We construct the score to be discretization invariant, which is natural given the underlying spatially continuous process. We conduct a series of experiments, ranging from synthetic examples with closed-form solutions to the stochastic nonlinear evolution of real-world biological shape data, and our method demonstrates high efficacy, particularly due to its ability to adapt to any resolution without extra training.
Authors: Dongjie Chen, Kartik Patwari, Zhengfeng Lai, Sen-ching Cheung, Chen-Nee Chuah
Abstract: Source-Free Domain Adaptation (SFDA) aims to adapt a pre-trained source model to a target domain using only unlabeled target data. Current SFDA methods face challenges in effectively leveraging pre-trained knowledge and exploiting target domain data. Multimodal Large Language Models (MLLMs) offer remarkable capabilities in understanding visual and textual information, but their applicability to SFDA poses challenges such as instruction-following failures, intensive computational demands, and difficulties in performance measurement prior to adaptation. To alleviate these issues, we propose Reliability-based Curriculum Learning (RCL), a novel framework that integrates multiple MLLMs for knowledge exploitation via pseudo-labeling in SFDA. Our framework incorporates proposed Reliable Knowledge Transfer, Self-correcting and MLLM-guided Knowledge Expansion, and Multi-hot Masking Refinement to progressively exploit unlabeled data in the target domain. RCL achieves state-of-the-art (SOTA) performance on multiple SFDA benchmarks, e.g., $\textbf{+9.4%}$ on DomainNet, demonstrating its effectiveness in enhancing adaptability and robustness without requiring access to source data. Code: https://github.com/Dong-Jie-Chen/RCL.
Authors: George Ma, Yifei Wang, Derek Lim, Stefanie Jegelka, Yisen Wang
Abstract: In many applications, we desire neural networks to exhibit invariance or equivariance to certain groups due to symmetries inherent in the data. Recently, frame-averaging methods emerged to be a unified framework for attaining symmetries efficiently by averaging over input-dependent subsets of the group, i.e., frames. What we currently lack is a principled understanding of the design of frames. In this work, we introduce a canonization perspective that provides an essential and complete view of the design of frames. Canonization is a classic approach for attaining invariance by mapping inputs to their canonical forms. We show that there exists an inherent connection between frames and canonical forms. Leveraging this connection, we can efficiently compare the complexity of frames as well as determine the optimality of certain frames. Guided by this principle, we design novel frames for eigenvectors that are strictly superior to existing methods -- some are even optimal -- both theoretically and empirically. The reduction to the canonization perspective further uncovers equivalences between previous methods. These observations suggest that canonization provides a fundamental understanding of existing frame-averaging methods and unifies existing equivariant and invariant learning methods.
Authors: Pengxiang Li, Lu Yin, Xiaowei Gao, Shiwei Liu
Abstract: The rapid advancements in Large Language Models (LLMs) have revolutionized various natural language processing tasks. However, the substantial size of LLMs presents significant challenges in training or fine-tuning. While parameter-efficient approaches such as low-rank adaptation (LoRA) have gained popularity, they often compromise performance compared to full-rank fine-tuning. In this paper, we propose Outlier-weighed Layerwise Sampled Low-Rank Projection (OwLore), a new memory-efficient fine-tuning approach, inspired by the layerwise outlier distribution of LLMs, which dynamically samples pre-trained layers to fine-tune instead of adding additional adaptors. We first interpret the outlier phenomenon through the lens of Heavy-Tailed Self-Regularization theory (HT-SR), discovering that layers with more outliers tend to be more heavy-tailed and consequently better trained. Inspired by this finding, OwLore strategically assigns higher sampling probabilities to layers with more outliers to better leverage the knowledge stored in pre-trained LLMs. To further mitigate the memory demands of fine-tuning, we integrate gradient low-rank projection into our approach, which facilitates each layer to be efficiently trained in a low-rank manner. By incorporating the efficient characteristics of low-rank and optimal layerwise sampling, OwLore significantly improves the memory-performance trade-off in LLM pruning. Our extensive experiments across various architectures, including LLaMa2, LLaMa3, and Mistral, demonstrate that OwLore consistently outperforms baseline approaches, including full fine-tuning. Specifically, it achieves up to a 1.1% average accuracy gain on the Commonsense Reasoning benchmark, a 3.0% improvement on MMLU, and a notable 10% boost on MT-Bench, while being more memory efficient. OwLore allows us to fine-tune LLaMa2-7B with only 21GB of memory.
Authors: Alexander H\"agele, Elie Bakouch, Atli Kosson, Loubna Ben Allal, Leandro Von Werra, Martin Jaggi
Abstract: Scale has become a main ingredient in obtaining strong machine learning models. As a result, understanding a model's scaling properties is key to effectively designing both the right training setup as well as future generations of architectures. In this work, we argue that scale and training research has been needlessly complex due to reliance on the cosine schedule, which prevents training across different lengths for the same model size. We investigate the training behavior of a direct alternative - constant learning rate and cooldowns - and find that it scales predictably and reliably similar to cosine. Additionally, we show that stochastic weight averaging yields improved performance along the training trajectory, without additional training costs, across different scales. Importantly, with these findings we demonstrate that scaling experiments can be performed with significantly reduced compute and GPU hours by utilizing fewer but reusable training runs.
Authors: Zhangyu Wang, Gengchen Mai, Krzysztof Janowicz, Ni Lao
Abstract: A wide range of (multivariate) temporal (1D) and spatial (2D) data analysis tasks, such as grouping vehicle sensor trajectories, can be formulated as clustering with given metric constraints. Existing metric-constrained clustering algorithms overlook the rich correlation between feature similarity and metric distance, i.e., metric autocorrelation. The model-based variations of these clustering algorithms (e.g. TICC and STICC) achieve SOTA performance, yet suffer from computational instability and complexity by using a metric-constrained Expectation-Maximization procedure. In order to address these two problems, we propose a novel clustering algorithm, MC-GTA (Model-based Clustering via Goodness-of-fit Tests with Autocorrelations). Its objective is only composed of pairwise weighted sums of feature similarity terms (square Wasserstein-2 distance) and metric autocorrelation terms (a novel multivariate generalization of classic semivariogram). We show that MC-GTA is effectively minimizing the total hinge loss for intra-cluster observation pairs not passing goodness-of-fit tests, i.e., statistically not originating from the same distribution. Experiments on 1D/2D synthetic and real-world datasets demonstrate that MC-GTA successfully incorporates metric autocorrelation. It outperforms strong baselines by large margins (up to 14.3% in ARI and 32.1% in NMI) with faster and stabler optimization (>10x speedup).
Authors: Erik Thordsen, Erich Schubert
Abstract: Many algorithms require discriminative boundaries, such as separating hyperplanes or hyperballs, or are specifically designed to work on spherical data. By applying inversive geometry, we show that the two discriminative boundaries can be used interchangeably, and that general Euclidean data can be transformed into spherical data, whenever a change in point distances is acceptable. We provide explicit formulae to embed general Euclidean data into spherical data and to unembed it back. We further show a duality between hyperspherical caps, i.e., the volume created by a separating hyperplane on spherical data, and hyperballs and provide explicit formulae to map between the two. We further provide equations to translate inner products and Euclidean distances between the two spaces, to avoid explicit embedding and unembedding. We also provide a method to enforce projections of the general Euclidean space onto hemi-hyperspheres and propose an intrinsic dimensionality based method to obtain "all-purpose" parameters. To show the usefulness of the cap-ball-duality, we discuss example applications in machine learning and vector similarity search.
Authors: Fu-Yun Wang, Zhaoyang Huang, Alexander William Bergman, Dazhong Shen, Peng Gao, Michael Lingelbach, Keqiang Sun, Weikang Bian, Guanglu Song, Yu Liu, Hongsheng Li, Xiaogang Wang
Abstract: The consistency model (CM) has recently made significant progress in accelerating the generation of diffusion models. However, its application to high-resolution, text-conditioned image generation in the latent space (a.k.a., LCM) remains unsatisfactory. In this paper, we identify three key flaws in the current design of LCM. We investigate the reasons behind these limitations and propose the Phased Consistency Model (PCM), which generalizes the design space and addresses all identified limitations. Our evaluations demonstrate that PCM significantly outperforms LCM across 1--16 step generation settings. While PCM is specifically designed for multi-step refinement, it achieves even superior or comparable 1-step generation results to previously state-of-the-art specifically designed 1-step methods. Furthermore, we show that PCM's methodology is versatile and applicable to video generation, enabling us to train the state-of-the-art few-step text-to-video generator. More details are available at https://g-u-n.github.io/projects/pcm/.
Authors: Nicklas Hansen, Jyothir S V, Vlad Sobal, Yann LeCun, Xiaolong Wang, Hao Su
Abstract: Whole-body control for humanoids is challenging due to the high-dimensional nature of the problem, coupled with the inherent instability of a bipedal morphology. Learning from visual observations further exacerbates this difficulty. In this work, we explore highly data-driven approaches to visual whole-body humanoid control based on reinforcement learning, without any simplifying assumptions, reward design, or skill primitives. Specifically, we propose a hierarchical world model in which a high-level agent generates commands based on visual observations for a low-level agent to execute, both of which are trained with rewards. Our approach produces highly performant control policies in 8 tasks with a simulated 56-DoF humanoid, while synthesizing motions that are broadly preferred by humans. Code and videos: https://nicklashansen.com/rlpuppeteer
Authors: Eliahu Horwitz, Asaf Shul, Yedid Hoshen
Abstract: The rapid growth of neural network models shared on the internet has made model weights an important data modality. However, this information is underutilized as the weights are uninterpretable, and publicly available models are disorganized. Inspired by Darwin's tree of life, we define the Model Tree which describes the origin of models i.e., the parent model that was used to fine-tune the target model. Similarly to the natural world, the tree structure is unknown. In this paper, we introduce the task of Model Tree Heritage Recovery (MoTHer Recovery) for discovering Model Trees in the ever-growing universe of neural networks. Our hypothesis is that model weights encode this information, the challenge is to decode the underlying tree structure given the weights. Beyond the immediate application of model authorship attribution, MoTHer recovery holds exciting long-term applications akin to indexing the internet by search engines. Practically, for each pair of models, this task requires: i) determining if they are related, and ii) establishing the direction of the relationship. We find that certain distributional properties of the weights evolve monotonically during training, which enables us to classify the relationship between two given models. MoTHer recovery reconstructs entire model hierarchies, represented by a directed tree, where a parent model gives rise to multiple child models through additional training. Our approach successfully reconstructs complex Model Trees, as well as the structure of "in-the-wild" model families such as Llama 2 and Stable Diffusion.
Authors: Xiaodong Liu
Abstract: This paper presents a significant improvement on the previous conference paper known as DefSent. The prior study seeks to improve sentence embeddings of language models by projecting definition sentences into the vector space of dictionary entries. We discover that this approach is not fully explored due to the methodological limitation of using word embeddings of language models to represent dictionary entries. This leads to two hindrances. First, dictionary entries are constrained by the single-word vocabulary, and thus cannot be fully exploited. Second, semantic representations of language models are known to be anisotropic, but pre-processing word embeddings for DefSent is not allowed because its weight is frozen during training and tied to the prediction layer. In this paper, we propose a novel method to progressively build entry embeddings not subject to the limitations. As a result, definition sentences can be projected into a quasi-isotropic or isotropic vector space of unlimited dictionary entries, so that sentence embeddings of noticeably better quality are attainable. We abbreviate our approach as DefSent+ (a plus version of DefSent), involving the following strengths: 1) the task performance on measuring sentence similarities is significantly improved compared to DefSent; 2) when DefSent+ is used to further train data-augmented models like SIMCSE and SNCSE, state-of-the-art performance on measuring sentence similarities can be achieved among the approaches without using manually labeled datasets; 3) DefSent+ is also competitive in feature-based transfer for NLP downstream tasks.
Authors: Chong Zheng, Yongming Huang, Cheng Zhang, Tony Q. S. Quek
Abstract: In this paper, we aim to maximize the SSR for heterogeneous service demands in the cooperative MEC-assisted RAN slicing system by jointly considering the multi-node computing resources cooperation and allocation, the transmission resource blocks (RBs) allocation, and the time-varying dynamicity of the system. To this end, we abstract the system into a weighted undirected topology graph and, then propose a recurrent graph reinforcement learning (RGRL) algorithm to intelligently learn the optimal hybrid RA policy. Therein, the graph neural network (GCN) and the deep deterministic policy gradient (DDPG) is combined to effectively extract spatial features from the equivalent topology graph. Furthermore, a novel time recurrent reinforcement learning framework is designed in the proposed RGRL algorithm by incorporating the action output of the policy network at the previous moment into the state input of the policy network at the subsequent moment, so as to cope with the time-varying and contextual network environment. In addition, we explore two use case scenarios to discuss the universal superiority of the proposed RGRL algorithm. Simulation results demonstrate the superiority of the proposed algorithm in terms of the average SSR, the performance stability, and the network complexity.
Authors: Simranjit Singh, Andreas Karatzas, Michael Fore, Iraklis Anagnostopoulos, Dimitrios Stamoulis
Abstract: State-of-the-art sequential reasoning in Large Language Models (LLMs) has expanded the capabilities of Copilots beyond conversational tasks to complex function calling, managing thousands of API calls. However, the tendency of compositional prompting to segment tasks into multiple steps, each requiring a round-trip to the GPT APIs, leads to increased system latency and costs. Although recent advancements in parallel function calling have improved tool execution per API call, they may necessitate more detailed in-context instructions and task breakdown at the prompt level, resulting in higher engineering and production costs. Inspired by the hardware design principles of multiply-add (MAD) operations, which fuse multiple arithmetic operations into a single task from the compiler's perspective, we propose LLM-Tool Compiler, which selectively fuses similar types of tool operations under a single function at runtime, presenting them as a unified task to the LLM. This selective fusion inherently enhances parallelization and efficiency. Benchmarked on a large-scale Copilot platform, LLM-Tool Compiler achieves up to four times more parallel calls than existing methods, reducing token costs and latency by up to 40% and 12%, respectively.
Authors: Hao Zhou, Chengming Hu, Xue Liu
Abstract: Reconfigurable intelligent surface (RIS) becomes a promising technique for 6G networks by reshaping signal propagation in smart radio environments. However, it also leads to significant complexity for network management due to the large number of elements and dedicated phase-shift optimization. In this work, we provide an overview of machine learning (ML)-enabled optimization for RIS-aided 6G networks. In particular, we focus on various reinforcement learning (RL) techniques, e.g., deep Q-learning, multi-agent reinforcement learning, transfer reinforcement learning, hierarchical reinforcement learning, and offline reinforcement learning. Different from existing studies, this work further discusses how large language models (LLMs) can be combined with RL to handle network optimization problems. It shows that LLM offers new opportunities to enhance the capabilities of RL algorithms in terms of generalization, reward function design, multi-modal information processing, etc. Finally, we identify the future challenges and directions of ML-enabled optimization for RIS-aided 6G networks.
Authors: Bhagyashri Tushir, Vikram K Ramanna, Yuhong Liu, Behnam Dezfouli
Abstract: Identifying IoT devices is crucial for network monitoring, security enforcement, and inventory tracking. However, most existing identification methods rely on deep packet inspection, which raises privacy concerns and adds computational complexity. More importantly, existing works overlook the impact of wireless channel dynamics on the accuracy of layer-2 features, thereby limiting their effectiveness in real-world scenarios. In this work, we define and use the latency of specific probe-response packet exchanges, referred to as "device latency," as the main feature for device identification. Additionally, we reveal the critical impact of wireless channel dynamics on the accuracy of device identification based on device latency. Specifically, this work introduces "accumulation score" as a novel approach to capturing fine-grained channel dynamics and their impact on device latency when training machine learning models. We implement the proposed methods and measure the accuracy and overhead of device identification in real-world scenarios. The results confirm that by incorporating the accumulation score for balanced data collection and training machine learning algorithms, we achieve an F1 score of over 97% for device identification, even amidst wireless channel dynamics, a significant improvement over the 75% F1 score achieved by disregarding the impact of channel dynamics on data collection and device latency.
Authors: Min Hun Lee
Abstract: In this paper, we explore the feasibility of using a transformer-based, spatiotemporal attention network (STAN) for gradient-based time-series explanations. First, we trained the STAN model for video classifications using the global and local views of data and weakly supervised labels on time-series data (i.e. the type of an activity). We then leveraged a gradient-based XAI technique (e.g. saliency map) to identify salient frames of time-series data. According to the experiments using the datasets of four medically relevant activities, the STAN model demonstrated its potential to identify important frames of videos.
Authors: Maximilian Mueller, Matthias Hein
Abstract: VisionTransformers have been shown to be powerful out-of-distribution detectors for ImageNet-scale settings when finetuned from publicly available checkpoints, often outperforming other model types on popular benchmarks. In this work, we investigate the impact of both the pretraining and finetuning scheme on the performance of ViTs on this task by analyzing a large pool of models. We find that the exact type of pretraining has a strong impact on which method works well and on OOD detection performance in general. We further show that certain training schemes might only be effective for a specific type of out-distribution, but not in general, and identify a best-practice training recipe.
Authors: Velmathi G, Shangavelan M, Harish D, Krithikshun M S
Abstract: This project undertakes the training and analysis of optical character recognition OCR methods applied to 10th century ancient Tamil inscriptions discovered on the walls of the Brihadeeswarar Temple.The chosen OCR methods include Tesseract,a widely used OCR engine,using modern ICR techniques to pre process the raw data and a box editing software to finetune our model.The analysis with Tesseract aims to evaluate their effectiveness in accurately deciphering the nuances of the ancient Tamil characters.The performance of our model for the dataset are determined by their accuracy rate where the evaluated dataset divided into training set and testing set.By addressing the unique challenges posed by the script's historical context,this study seeks to contribute valuable insights to the broader field of OCR,facilitating improved preservation and interpretation of ancient inscriptions
Authors: Thomas Winterbottom, G. Thomas Hudson, Daniel Kluvanec, Dean Slack, Jamie Sterling, Junjie Shentu, Chenghao Xiao, Zheming Zhou, Noura Al Moubayed
Abstract: Next-frame prediction is a useful and powerful method for modelling and understanding the dynamics of video data. Inspired by the empirical success of causal language modelling and next-token prediction in language modelling, we explore the extent to which next-frame prediction serves as a strong foundational learning strategy (analogous to language modelling) for inducing an understanding of the visual world. In order to quantify the specific visual understanding induced by next-frame prediction, we introduce six diagnostic simulation video datasets derived from fundamental physical laws created by varying physical constants such as gravity and mass. We demonstrate that our models trained only on next-frame prediction are capable of predicting the value of these physical constants (e.g. gravity) without having been trained directly to learn these constants via a regression task. We find that the generative training phase alone induces a model state that can predict physical constants significantly better than that of a random model, improving the loss by a factor of between 1.28 to 6.24. We conclude that next-frame prediction shows great promise as a general learning strategy to induce understanding of the many `laws' that govern the visual domain without the need for explicit labelling.
Authors: Adib Hasan, Mardavij Roozbehani, Munther Dahleh
Abstract: This paper introduces WeatherFormer, a transformer encoder-based model designed to learn robust weather features from minimal observations. It addresses the challenge of modeling complex weather dynamics from small datasets, a bottleneck for many prediction tasks in agriculture, epidemiology, and climate science. WeatherFormer was pretrained on a large pretraining dataset comprised of 39 years of satellite measurements across the Americas. With a novel pretraining task and fine-tuning, WeatherFormer achieves state-of-the-art performance in county-level soybean yield prediction and influenza forecasting. Technical innovations include a unique spatiotemporal encoding that captures geographical, annual, and seasonal variations, adapting the transformer architecture to continuous weather data, and a pretraining strategy to learn representations that are robust to missing weather features. This paper for the first time demonstrates the effectiveness of pretraining large transformer encoder models for weather-dependent applications across multiple domains.
Authors: Ling-Qi Zhang, Zahra Kadkhodaie, Eero P. Simoncelli, David H. Brainard
Abstract: We re-examine the problem of reconstructing a high-dimensional signal from a small set of linear measurements, in combination with image prior from a diffusion probabilistic model. Well-established methods for optimizing such measurements include principal component analysis (PCA), independent component analysis (ICA) and compressed sensing (CS), all of which rely on axis- or subspace-aligned statistical characterization. But many naturally occurring signals, including photographic images, contain richer statistical structure. To exploit such structure, we introduce a general method for obtaining an optimized set of linear measurements, assuming a Bayesian inverse solution that leverages the prior implicit in a neural network trained to perform denoising. We demonstrate that these measurements are distinct from those of PCA and CS, with significant improvements in minimizing squared reconstruction error. In addition, we show that optimizing the measurements for the SSIM perceptual loss leads to perceptually improved reconstruction. Our results highlight the importance of incorporating the specific statistical regularities of natural signals when designing effective linear measurements.
Authors: Naibo Wang, Yuchen Deng, Wenjie Feng, Jianwei Yin, See-Kiong Ng
Abstract: Federated Class Incremental Learning (FCIL) is a critical yet largely underexplored issue that deals with the dynamic incorporation of new classes within federated learning (FL). Existing methods often employ generative adversarial networks (GANs) to produce synthetic images to address privacy concerns in FL. However, GANs exhibit inherent instability and high sensitivity, compromising the effectiveness of these methods. In this paper, we introduce a novel data-free federated class incremental learning framework with diffusion-based generative memory (DFedDGM) to mitigate catastrophic forgetting by generating stable, high-quality images through diffusion models. We design a new balanced sampler to help train the diffusion models to alleviate the common non-IID problem in FL, and introduce an entropy-based sample filtering technique from an information theory perspective to enhance the quality of generative samples. Finally, we integrate knowledge distillation with a feature-based regularization term for better knowledge transfer. Our framework does not incur additional communication costs compared to the baseline FedAvg method. Extensive experiments across multiple datasets demonstrate that our method significantly outperforms existing baselines, e.g., over a 4% improvement in average accuracy on the Tiny-ImageNet dataset.
Authors: Ningyuan Chen, Xuefeng Gao, Yi Xiong
Abstract: When two players are engaged in a repeated game with unknown payoff matrices, they may be completely unaware of the existence of each other and use multi-armed bandit algorithms to choose the actions, which is referred to as the ``blindfolded game'' in this paper. We show that when the players use Thompson sampling, the game dynamics converges to the Nash equilibrium under a mild assumption on the payoff matrices. Therefore, algorithmic collusion doesn't arise in this case despite the fact that the players do not intentionally deploy competitive strategies. To prove the convergence result, we find that the framework developed in stochastic approximation doesn't apply, because of the sporadic and infrequent updates of the inferior actions and the lack of Lipschitz continuity. We develop a novel sample-path-wise approach to show the convergence.
Authors: Olena Burda-Lassen, Aman Chadha, Shashank Goswami, Vinija Jain
Abstract: An image is often said to be worth a thousand words, and certain images can tell rich and insightful stories. Can these stories be told via image captioning? Images from folklore genres, such as mythology, folk dance, cultural signs, and symbols, are vital to every culture. Our research compares the performance of four popular vision-language models (GPT-4V, Gemini Pro Vision, LLaVA, and OpenFlamingo) in identifying culturally specific information in such images and creating accurate and culturally sensitive image captions. We also propose a new evaluation metric, Cultural Awareness Score (CAS), dedicated to measuring the degree of cultural awareness in image captions. We provide a dataset MOSAIC-1.5k, labeled with ground truth for images containing cultural background and context, as well as a labeled dataset with assigned Cultural Awareness Scores that can be used with unseen data. Creating culturally appropriate image captions is valuable for scientific research and can be beneficial for many practical applications. We envision that our work will promote a deeper integration of cultural sensitivity in AI applications worldwide. By making the dataset and Cultural Awareness Score available to the public, we aim to facilitate further research in this area, encouraging the development of more culturally aware AI systems that respect and celebrate global diversity.
Authors: Rinat I. Dumaev, Sergei A. Molodyakov, Lev V. Utkin
Abstract: To increase the transparency of modern computer-aided diagnosis (CAD) systems for assessing the malignancy of lung nodules, an interpretable model based on applying the generalized additive models and the concept-based learning is proposed. The model detects a set of clinically significant attributes in addition to the final malignancy regression score and learns the association between the lung nodule attributes and a final diagnosis decision as well as their contributions into the decision. The proposed concept-based learning framework provides human-readable explanations in terms of different concepts (numerical and categorical), their values, and their contribution to the final prediction. Numerical experiments with the LIDC-IDRI dataset demonstrate that the diagnosis results obtained using the proposed model, which explicitly explores internal relationships, are in line with similar patterns observed in clinical practice. Additionally, the proposed model shows the competitive classification and the nodule attribute scoring performance, highlighting its potential for effective decision-making in the lung nodule diagnosis.
Authors: Alexander DeRieux, Walid Saad
Abstract: Collaboration is a key challenge in distributed multi-agent reinforcement learning (MARL) environments. Learning frameworks for these decentralized systems must weigh the benefits of explicit player coordination against the communication overhead and computational cost of sharing local observations and environmental data. Quantum computing has sparked a potential synergy between quantum entanglement and cooperation in multi-agent environments, which could enable more efficient distributed collaboration with minimal information sharing. This relationship is largely unexplored, however, as current state-of-the-art quantum MARL (QMARL) implementations rely on classical information sharing rather than entanglement over a quantum channel as a coordination medium. In contrast, in this paper, a novel framework dubbed entangled QMARL (eQMARL) is proposed. The proposed eQMARL is a distributed actor-critic framework that facilitates cooperation over a quantum channel and eliminates local observation sharing via a quantum entangled split critic. Introducing a quantum critic uniquely spread across the agents allows coupling of local observation encoders through entangled input qubits over a quantum channel, which requires no explicit sharing of local observations and reduces classical communication overhead. Further, agent policies are tuned through joint observation-value function estimation via joint quantum measurements, thereby reducing the centralized computational burden. Experimental results show that eQMARL with ${\Psi}^{+}$ entanglement converges to a cooperative strategy up to $17.8\%$ faster and with a higher overall score compared to split classical and fully centralized classical and quantum baselines. The results also show that eQMARL achieves this performance with a constant factor of $25$-times fewer centralized parameters compared to the split classical baseline.
Authors: Pierre-Fran\c{c}ois Marteau (EXPRESSION)
Abstract: We introduce and detail an atypical neural network architecture, called time elastic neural network (teNN), for multivariate time series classification. The novelty compared to classical neural network architecture is that it explicitly incorporates time warping ability, as well as a new way of considering attention. In addition, this architecture is capable of learning a dropout strategy, thus optimizing its own architecture.Behind the design of this architecture, our overall objective is threefold: firstly, we are aiming at improving the accuracy of instance based classification approaches that shows quite good performances as far as enough training data is available. Secondly we seek to reduce the computational complexity inherent to these methods to improve their scalability. Ideally, we seek to find an acceptable balance between these first two criteria. And finally, we seek to enhance the explainability of the decision provided by this kind of neural architecture.The experiment demonstrates that the stochastic gradient descent implemented to train a teNN is quite effective. To the extent that the selection of some critical meta-parameters is correct, convergence is generally smooth and fast.While maintaining good accuracy, we get a drastic gain in scalability by first reducing the required number of reference time series, i.e. the number of teNN cells required. Secondly, we demonstrate that, during the training process, the teNN succeeds in reducing the number of neurons required within each cell. Finally, we show that the analysis of the activation and attention matrices as well as the reference time series after training provides relevant information to interpret and explain the classification results.The comparative study that we have carried out and which concerns around thirty diverse and multivariate datasets shows that the teNN obtains results comparable to those of the state of the art, in particular similar to those of a network mixing LSTM and CNN architectures for example.
Authors: Franz Motzkus, Christian Hellert, Ute Schmid
Abstract: Ensuring the quality of black-box Deep Neural Networks (DNNs) has become ever more significant, especially in safety-critical domains such as automated driving. While global concept encodings generally enable a user to test a model for a specific concept, linking global concept encodings to the local processing of single network inputs reveals their strengths and limitations. Our proposed framework global-to-local Concept Attribution (glCA) uses approaches from local (why a specific prediction originates) and global (how a model works generally) eXplainable Artificial Intelligence (xAI) to test DNNs for a predefined semantical concept locally. The approach allows for conditioning local, post-hoc explanations on predefined semantic concepts encoded as linear directions in the model's latent space. Pixel-exact scoring concerning the global concept usage assists the tester in further understanding the model processing of single data points for the selected concept. Our approach has the advantage of fully covering the model-internal encoding of the semantic concept and allowing the localization of relevant concept-related information. The results show major differences in the local perception and usage of individual global concept encodings and demand for further investigations regarding obtaining thorough semantic concept encodings.
Authors: Apurva Sinha, Ekta Gujral
Abstract: Product attribute extraction is an growing field in e-commerce business, with several applications including product ranking, product recommendation, future assortment planning and improving online shopping customer experiences. Understanding the customer needs is critical part of online business, specifically fashion products. Retailers uses assortment planning to determine the mix of products to offer in each store and channel, stay responsive to market dynamics and to manage inventory and catalogs. The goal is to offer the right styles, in the right sizes and colors, through the right channels. When shoppers find products that meet their needs and desires, they are more likely to return for future purchases, fostering customer loyalty. Product attributes are a key factor in assortment planning. In this paper we present PAE, a product attribute extraction algorithm for future trend reports consisting text and images in PDF format. Most existing methods focus on attribute extraction from titles or product descriptions or utilize visual information from existing product images. Compared to the prior works, our work focuses on attribute extraction from PDF files where upcoming fashion trends are explained. This work proposes a more comprehensive framework that fully utilizes the different modalities for attribute extraction and help retailers to plan the assortment in advance. Our contributions are three-fold: (a) We develop PAE, an efficient framework to extract attributes from unstructured data (text and images); (b) We provide catalog matching methodology based on BERT representations to discover the existing attributes using upcoming attribute values; (c) We conduct extensive experiments with several baselines and show that PAE is an effective, flexible and on par or superior (avg 92.5% F1-Score) framework to existing state-of-the-art for attribute value extraction task.
Authors: Jessica N. Howard, Marc S. Klinger, Anindita Maiti, Alexander G. Stapleton
Abstract: The Neural Network Field Theory correspondence (NNFT) is a mapping from neural network (NN) architectures into the space of statistical field theories (SFTs). The Bayesian renormalization group (BRG) is an information-theoretic coarse graining scheme that generalizes the principles of the Exact Renormalization Group (ERG) to arbitrarily parameterized probability distributions, including those of NNs. In BRG, coarse graining is performed in parameter space with respect to an information-theoretic distinguishability scale set by the Fisher information metric. In this paper, we unify NNFT and BRG to form a powerful new framework for exploring the space of NNs and SFTs, which we coin BRG-NNFT. With BRG-NNFT, NN training dynamics can be interpreted as inducing a flow in the space of SFTs from the information-theoretic `IR' $\rightarrow$ `UV'. Conversely, applying an information-shell coarse graining to the trained network's parameters induces a flow in the space of SFTs from the information-theoretic `UV' $\rightarrow$ `IR'. When the information-theoretic cutoff scale coincides with a standard momentum scale, BRG is equivalent to ERG. We demonstrate the BRG-NNFT correspondence on two analytically tractable examples. First, we construct BRG flows for trained, infinite-width NNs, of arbitrary depth, with generic activation functions. As a special case, we then restrict to architectures with a single infinitely-wide layer, scalar outputs, and generalized cos-net activations. In this case, we show that BRG coarse-graining corresponds exactly to the momentum-shell ERG flow of a free scalar SFT. Our analytic results are corroborated by a numerical experiment in which an ensemble of asymptotically wide NNs are trained and subsequently renormalized using an information-shell BRG scheme.
Authors: Dominik S. Kufel, Jack Kemp, Simon M. Linsel, Chris R. Laumann, Norman Y. Yao
Abstract: We propose and analyze a family of approximately-symmetric neural networks for quantum spin liquid problems. These tailored architectures are parameter-efficient, scalable, and significantly out-perform existing symmetry-unaware neural network architectures. Utilizing the mixed-field toric code model, we demonstrate that our approach is competitive with the state-of-the-art tensor network and quantum Monte Carlo methods. Moreover, at the largest system sizes (N=480), our method allows us to explore Hamiltonians with sign problems beyond the reach of both quantum Monte Carlo and finite-size matrix-product states. The network comprises an exactly symmetric block following a non-symmetric block, which we argue learns a transformation of the ground state analogous to quasiadiabatic continuation. Our work paves the way toward investigating quantum spin liquid problems within interpretable neural network architectures
Authors: David Harvey
Abstract: Different models of dark matter can alter the distribution of mass in galaxy clusters in a variety of ways. However, so can uncertain astrophysical feedback mechanisms. Here we present a Machine Learning method that ''learns'' how the impact of dark matter self-interactions differs from that of astrophysical feedback in order to break this degeneracy and make inferences on dark matter. We train a Convolutional Neural Network on images of galaxy clusters from hydro-dynamic simulations. In the idealised case our algorithm is 80% accurate at identifying if a galaxy cluster harbours collisionless dark matter, dark matter with ${\sigma}_{\rm DM}/m = 0.1$cm$^2/$g or with ${\sigma}_{DM}/m = 1$cm$^2$/g. Whilst we find adding X-ray emissivity maps does not improve the performance in differentiating collisional dark matter, it does improve the ability to disentangle different models of astrophysical feedback. We include noise to resemble data expected from Euclid and Chandra and find our model has a statistical error of < 0.01cm$^2$/g and that our algorithm is insensitive to shape measurement bias and photometric redshift errors. This method represents a new way to analyse data from upcoming telescopes that is an order of magnitude more precise and many orders faster, enabling us to explore the dark matter parameter space like never before.
Authors: Arthur Jacot, Alexandre Kaiser
Abstract: We study Leaky ResNets, which interpolate between ResNets ($\tilde{L}=0$) and Fully-Connected nets ($\tilde{L}\to\infty$) depending on an 'effective depth' hyper-parameter $\tilde{L}$. In the infinite depth limit, we study 'representation geodesics' $A_{p}$: continuous paths in representation space (similar to NeuralODEs) from input $p=0$ to output $p=1$ that minimize the parameter norm of the network. We give a Lagrangian and Hamiltonian reformulation, which highlight the importance of two terms: a kinetic energy which favors small layer derivatives $\partial_{p}A_{p}$ and a potential energy that favors low-dimensional representations, as measured by the 'Cost of Identity'. The balance between these two forces offers an intuitive understanding of feature learning in ResNets. We leverage this intuition to explain the emergence of a bottleneck structure, as observed in previous work: for large $\tilde{L}$ the potential energy dominates and leads to a separation of timescales, where the representation jumps rapidly from the high dimensional inputs to a low-dimensional representation, move slowly inside the space of low-dimensional representations, before jumping back to the potentially high-dimensional outputs. Inspired by this phenomenon, we train with an adaptive layer step-size to adapt to the separation of timescales.
Authors: Emile Contal, Garrin McGoldrick
Abstract: Large Language Models (LLM) hold immense promise for real-world applications, but their generic knowledge often falls short of domain-specific needs. Fine-tuning, a common approach, can suffer from catastrophic forgetting and hinder generalizability. In-Context Learning (ICL) offers an alternative, which can leverage Retrieval-Augmented Generation (RAG) to provide LLMs with relevant demonstrations for few-shot learning tasks. This paper explores the desired qualities of a demonstration retrieval system for ICL. We argue that ICL retrieval in this context resembles item-cold-start recommender systems, prioritizing discovery and maximizing information gain over strict relevance. We propose a novel evaluation method that measures the LLM's subsequent performance on NLP tasks, eliminating the need for subjective diversity scores. Our findings demonstrate the critical role of diversity and quality bias in retrieved demonstrations for effective ICL, and highlight the potential of recommender system techniques in this domain.
Authors: Armin Moradi, Nicola Neophytou, Golnoosh Farnadi
Abstract: Popularity bias in music recommendation systems -- where artists and tracks with the highest listen counts are recommended more often -- can also propagate biases along demographic and cultural axes. In this work, we identify these biases in recommendations for artists from underrepresented cultural groups in prototype-based matrix factorization methods. Unlike traditional matrix factorization methods, prototype-based approaches are interpretable. This allows us to directly link the observed bias in recommendations for minority artists (the effect) to specific properties of the embedding space (the cause). We mitigate popularity bias in music recommendation through capturing both users' and songs' cultural nuances in the embedding space. To address these challenges while maintaining recommendation quality, we propose two novel enhancements to the embedding space: i) we propose an approach to filter-out the irrelevant prototypes used to represent each user and item to improve generalizability, and ii) we introduce regularization techniques to reinforce a more uniform distribution of prototypes within the embedding space. Our results demonstrate significant improvements in reducing popularity bias and enhancing demographic and cultural fairness in music recommendations while achieving competitive -- if not better -- overall performance.
Authors: Francisco de Arriba-P\'erez, Silvia Garc\'ia-M\'endez, Francisco J. Gonz\'alez-Casta\~no, Jaime Gonz\'alez-Gonz\'alez
Abstract: Artificial Intelligence techniques such as Machine Learning (ML) have not been exploited to their maximum potential in the legal domain. This has been partially due to the insufficient explanations they provided about their decisions. Automatic expert systems with explanatory capabilities can be specially useful when legal practitioners search jurisprudence to gather contextual knowledge for their cases. Therefore, we propose a hybrid system that applies ML for multi-label classification of judgements (sentences) and visual and natural language descriptions for explanation purposes, boosted by Natural Language Processing techniques and deep legal reasoning to identify the entities, such as the parties, involved. We are not aware of any prior work on automatic multi-label classification of legal judgements also providing natural language explanations to the end-users with comparable overall quality. Our solution achieves over 85 % micro precision on a labelled data set annotated by legal experts. This endorses its interest to relieve human experts from monotonous labour-intensive legal classification tasks.
Authors: Dibyajyoti Chakraborty, Shivam Barwey, Hong Zhang, Romit Maulik
Abstract: In this work, we provide a mathematical formulation for error propagation in flow trajectory prediction using data-driven turbulence closure modeling. Under the assumption that the predicted state of a large eddy simulation prediction must be close to that of a subsampled direct numerical simulation, we retrieve an upper bound for the prediction error when utilizing a data-driven closure model. We also demonstrate that this error is significantly affected by the time step size and the Jacobian which play a role in amplifying the initial one-step error made by using the closure. Our analysis also shows that the error propagates exponentially with rollout time and the upper bound of the system Jacobian which is itself influenced by the Jacobian of the closure formulation. These findings could enable the development of new regularization techniques for ML models based on the identified error-bound terms, improving their robustness and reducing error propagation.
Authors: Divyam Madaan, Taro Makino, Sumit Chopra, Kyunghyun Cho
Abstract: Supervised multi-modal learning involves mapping multiple modalities to a target label. Previous studies in this field have concentrated on capturing in isolation either the inter-modality dependencies (the relationships between different modalities and the label) or the intra-modality dependencies (the relationships within a single modality and the label). We argue that these conventional approaches that rely solely on either inter- or intra-modality dependencies may not be optimal in general. We view the multi-modal learning problem from the lens of generative models where we consider the target as a source of multiple modalities and the interaction between them. Towards that end, we propose inter- & intra-modality modeling (I2M2) framework, which captures and integrates both the inter- and intra-modality dependencies, leading to more accurate predictions. We evaluate our approach using real-world healthcare and vision-and-language datasets with state-of-the-art models, demonstrating superior performance over traditional methods focusing only on one type of modality dependency.
Authors: Francesco Paissan, Luca Della Libera, Mirco Ravanelli, Cem Subakan
Abstract: Interpreting the decisions of deep learning models, including audio classifiers, is crucial for ensuring the transparency and trustworthiness of this technology. In this paper, we introduce LMAC-ZS (Listenable Maps for Audio Classifiers in the Zero-Shot context), which, to the best of our knowledge, is the first decoder-based post-hoc interpretation method for explaining the decisions of zero-shot audio classifiers. The proposed method utilizes a novel loss function that maximizes the faithfulness to the original similarity between a given text-and-audio pair. We provide an extensive evaluation using the Contrastive Language-Audio Pretraining (CLAP) model to showcase that our interpreter remains faithful to the decisions in a zero-shot classification context. Moreover, we qualitatively show that our method produces meaningful explanations that correlate well with different text prompts.
Authors: Tommy Rochussen
Abstract: Bayesian neural network posterior distributions have a great number of modes that correspond to the same network function. The abundance of such modes can make it difficult for approximate inference methods to do their job. Recent work has demonstrated the benefits of partial stochasticity for approximate inference in Bayesian neural networks; inference can be less costly and performance can sometimes be improved. I propose a structured way to select the deterministic subset of weights that removes neuron permutation symmetries, and therefore the corresponding redundant posterior modes. With a drastically simplified posterior distribution, the performance of existing approximate inference schemes is found to be greatly improved.
Authors: Malia L. Kao, Keith Hawkins, Laura K. Rogers, Amy Bonsor, Bart H. Dunlap, Jason L. Sanders, M. H. Montgomery, D. E. Winget
Abstract: White dwarfs (WDs) polluted by exoplanetary material provide the unprecedented opportunity to directly observe the interiors of exoplanets. However, spectroscopic surveys are often limited by brightness constraints, and WDs tend to be very faint, making detections of large populations of polluted WDs difficult. In this paper, we aim to increase considerably the number of WDs with multiple metals in their atmospheres. Using 96,134 WDs with Gaia DR3 BP/RP (XP) spectra, we constructed a 2D map using an unsupervised machine learning technique called Uniform Manifold Approximation and Projection (UMAP) to organize the WDs into identifiable spectral regions. The polluted WDs are among the distinct spectral groups identified in our map. We have shown that this selection method could potentially increase the number of known WDs with 5 or more metal species in their atmospheres by an order of magnitude. Such systems are essential for characterizing exoplanet diversity and geology.
Authors: Kushagra Pandey, Ruihan Yang, Stephan Mandt
Abstract: Constructing fast samplers for unconditional diffusion and flow-matching models has received much attention recently; however, existing methods for solving inverse problems, such as super-resolution, inpainting, or deblurring, still require hundreds to thousands of iterative steps to obtain high-quality results. We propose a plug-and-play framework for constructing efficient samplers for inverse problems, requiring only pre-trained diffusion or flow-matching models. We present Conditional Conjugate Integrators, which leverage the specific form of the inverse problem to project the respective conditional diffusion/flow dynamics into a more amenable space for sampling. Our method complements popular posterior approximation methods for solving inverse problems using diffusion/flow models. We evaluate the proposed method's performance on various linear image restoration tasks across multiple datasets, employing diffusion and flow-matching models. Notably, on challenging inverse problems like 4$\times$ super-resolution on the ImageNet dataset, our method can generate high-quality samples in as few as 5 conditional sampling steps and outperforms competing baselines requiring 20-1000 steps. Our code and models will be publicly available at https://github.com/mandt-lab/CI2RM.
Authors: Saeedeh Ghanadbashi, Fatemeh Golpayegani
Abstract: Agents, whether software or hardware, perceive their environment through sensors and act using actuators, often operating in dynamic, partially observable settings. They face challenges like incomplete and noisy data, unforeseen situations, and the need to adapt goals in real-time. Traditional reasoning and ML methods, including Reinforcement Learning (RL), help but are limited by data needs, predefined goals, and extensive exploration periods. Ontologies offer a solution by integrating diverse information sources, enhancing decision-making in complex environments. This thesis introduces an ontology-enhanced decision-making model (OntoDeM) for autonomous agents. OntoDeM enriches agents' domain knowledge, allowing them to interpret unforeseen events, generate or adapt goals, and make better decisions. Key contributions include: 1. An ontology-based method to improve agents' real-time observations using prior knowledge. 2. The OntoDeM model for handling dynamic, unforeseen situations by evolving or generating new goals. 3. Implementation and evaluation in four real-world applications, demonstrating its effectiveness. Compared to traditional and advanced learning algorithms, OntoDeM shows superior performance in improving agents' observations and decision-making in dynamic, partially observable environments.
Authors: Iosif Lytras, Panayotis Mertikopoulos
Abstract: Motivated by applications to deep learning which often fail standard Lipschitz smoothness requirements, we examine the problem of sampling from distributions that are not log-concave and are only weakly dissipative, with log-gradients allowed to grow superlinearly at infinity. In terms of structure, we only assume that the target distribution satisfies either a log-Sobolev or a Poincar\'e inequality and a local Lipschitz smoothness assumption with modulus growing possibly polynomially at infinity. This set of assumptions greatly exceeds the operational limits of the "vanilla" unadjusted Langevin algorithm (ULA), making sampling from such distributions a highly involved affair. To account for this, we introduce a taming scheme which is tailored to the growth and decay properties of the target distribution, and we provide explicit non-asymptotic guarantees for the proposed sampler in terms of the Kullback-Leibler (KL) divergence, total variation, and Wasserstein distance to the target distribution.
Authors: Kanad Shrikar Pardeshi, Itai Shapira, Ariel D. Procaccia, Aarti Singh
Abstract: Is it possible to understand or imitate a policy maker's rationale by looking at past decisions they made? We formalize this question as the problem of learning social welfare functions belonging to the well-studied family of power mean functions. We focus on two learning tasks; in the first, the input is vectors of utilities of an action (decision or policy) for individuals in a group and their associated social welfare as judged by a policy maker, whereas in the second, the input is pairwise comparisons between the welfares associated with a given pair of utility vectors. We show that power mean functions are learnable with polynomial sample complexity in both cases, even if the comparisons are social welfare information is noisy. Finally, we design practical algorithms for these tasks and evaluate their performance.
Authors: Ahatsham Hayat, Mohammad Rashedul Hasan
Abstract: This paper introduces the Contextual Language model for Accurate Imputation Method (CLAIM), a novel strategy that capitalizes on the expansive knowledge and reasoning capabilities of pre-trained large language models (LLMs) to address missing data challenges in tabular datasets. Unlike traditional imputation methods, which predominantly rely on numerical estimations, CLAIM utilizes contextually relevant natural language descriptors to fill missing values. This approach transforms datasets into natural language contextualized formats that are inherently more aligned with LLMs' capabilities, thereby facilitating the dual use of LLMs: first, to generate missing value descriptors, and then, to fine-tune the LLM on the enriched dataset for improved performance in downstream tasks. Our evaluations across diverse datasets and missingness patterns reveal CLAIM's superior performance over existing imputation techniques. Furthermore, our investigation into the effectiveness of context-specific versus generic descriptors for missing data highlights the importance of contextual accuracy in enhancing LLM performance for data imputation. The results underscore CLAIM's potential to markedly improve the reliability and quality of data analysis and machine learning models, offering a more nuanced and effective solution for handling missing data.
Authors: Micah Carroll, Davis Foote, Anand Siththaranjan, Stuart Russell, Anca Dragan
Abstract: Existing AI alignment approaches assume that preferences are static, which is unrealistic: our preferences change, and may even be influenced by our interactions with AI systems themselves. To clarify the consequences of incorrectly assuming static preferences, we introduce Dynamic Reward Markov Decision Processes (DR-MDPs), which explicitly model preference changes and the AI's influence on them. We show that despite its convenience, the static-preference assumption may undermine the soundness of existing alignment techniques, leading them to implicitly reward AI systems for influencing user preferences in ways users may not truly want. We then explore potential solutions. First, we offer a unifying perspective on how an agent's optimization horizon may partially help reduce undesirable AI influence. Then, we formalize different notions of AI alignment that account for preference change from the outset. Comparing the strengths and limitations of 8 such notions of alignment, we find that they all either err towards causing undesirable AI influence, or are overly risk-averse, suggesting that a straightforward solution to the problems of changing preferences may not exist. As there is no avoiding grappling with changing preferences in real-world settings, this makes it all the more important to handle these issues with care, balancing risks and capabilities. We hope our work can provide conceptual clarity and constitute a first step towards AI alignment practices which explicitly account for (and contend with) the changing and influenceable nature of human preferences.
Authors: Sihe Zhang, Qingdong He, Jinlong Peng, Yuxi Li, Zhengkai Jiang, Jiafu Wu, Mingmin Chi, Yabiao Wang, Chengjie Wang
Abstract: Image retrieval aims to identify visually similar images within a database using a given query image. Traditional methods typically employ both global and local features extracted from images for matching, and may also apply re-ranking techniques to enhance accuracy. However, these methods often fail to account for the noise present in query images, which can stem from natural or human-induced factors, thereby negatively impacting retrieval performance. To mitigate this issue, we introduce a novel setting for low-quality image retrieval, and propose an Adaptive Noise-Based Network (AdapNet) to learn robust abstract representations. Specifically, we devise a quality compensation block trained to compensate for various low-quality factors in input images. Besides, we introduce an innovative adaptive noise-based loss function, which dynamically adjusts its focus on the gradient in accordance with image quality, thereby augmenting the learning of unknown noisy samples during training and enhancing intra-class compactness. To assess the performance, we construct two datasets with low-quality queries, which is built by applying various types of noise on clean query images on the standard Revisited Oxford and Revisited Paris datasets. Comprehensive experimental results illustrate that AdapNet surpasses state-of-the-art methods on the Noise Revisited Oxford and Noise Revisited Paris benchmarks, while maintaining competitive performance on high-quality datasets. The code and constructed datasets will be made available.
Authors: Inhwa Han, Jaayeon Lee, Jong Chul Ye
Abstract: Research efforts to understand neural signals have been ongoing for many years, with visual decoding from fMRI signals attracting considerable attention. Particularly, the advent of image diffusion models has advanced the reconstruction of images from fMRI data significantly. However, existing approaches often introduce inter- and intra- subject variations in the reconstructed images, which can compromise accuracy. To address current limitations in multi-subject brain decoding, we introduce a new Transformer architecture called MindFormer. This model is specifically designed to generate fMRI-conditioned feature vectors that can be used for conditioning Stable Diffusion model. More specifically, MindFormer incorporates two key innovations: 1) a novel training strategy based on the IP-Adapter to extract semantically meaningful features from fMRI signals, and 2) a subject specific token and linear layer that effectively capture individual differences in fMRI signals while synergistically combines multi subject fMRI data for training. Our experimental results demonstrate that Stable Diffusion, when integrated with MindFormer, produces semantically consistent images across different subjects. This capability significantly surpasses existing models in multi-subject brain decoding. Such advancements not only improve the accuracy of our reconstructions but also deepen our understanding of neural processing variations among individuals.
Authors: Yake Wei, Di Hu
Abstract: Multimodal learning methods with targeted unimodal learning objectives have exhibited their superior efficacy in alleviating the imbalanced multimodal learning problem. However, in this paper, we identify the previously ignored gradient conflict between multimodal and unimodal learning objectives, potentially misleading the unimodal encoder optimization. To well diminish these conflicts, we observe the discrepancy between multimodal loss and unimodal loss, where both gradient magnitude and covariance of the easier-to-learn multimodal loss are smaller than the unimodal one. With this property, we analyze Pareto integration under our multimodal scenario and propose MMPareto algorithm, which could ensure a final gradient with direction that is common to all learning objectives and enhanced magnitude to improve generalization, providing innocent unimodal assistance. Finally, experiments across multiple types of modalities and frameworks with dense cross-modal interaction indicate our superior and extendable method performance. Our method is also expected to facilitate multi-task cases with a clear discrepancy in task difficulty, demonstrating its ideal scalability. The source code and dataset are available at https://github.com/GeWu-Lab/MMPareto_ICML2024.
Authors: Zhengyang Tang, Chenyu Huang, Xin Zheng, Shixi Hu, Zizhuo Wang, Dongdong Ge, Benyou Wang
Abstract: Large Language Models (LLMs) have emerged as powerful tools for complex Operations Research (OR) in automating optimization modeling. However, current methodologies heavily rely on prompt engineering (e.g., multi-agent cooperation) with proprietary LLMs, raising data privacy concerns that could be prohibitive in industry applications. To tackle this issue, we propose training open-source LLMs for optimization modeling. We identify four critical requirements for the training dataset of OR LLMs, design and implement OR-Instruct, a semi-automated process for creating synthetic data tailored to specific requirements. We also introduce the IndustryOR benchmark, the first industrial benchmark for testing LLMs on solving real-world OR problems. We apply the data from OR-Instruct to various open-source LLMs of 7b size (termed as ORLMs), resulting in a significantly improved capability for optimization modeling. Our best-performing ORLM achieves state-of-the-art performance on the NL4OPT, MAMO, and IndustryOR benchmarks. Our code and data will be available at \url{https://github.com/Cardinal-Operations/ORLM}.
Authors: Zhifeng Chen, Kamlesh Pawar, Kh Tohidul Islam, Himashi Peiris, Gary Egan, Zhaolin Chen
Abstract: Motion artifacts in Magnetic Resonance Imaging (MRI) are one of the frequently occurring artifacts due to patient movements during scanning. Motion is estimated to be present in approximately 30% of clinical MRI scans; however, motion has not been explicitly modeled within deep learning image reconstruction models. Deep learning (DL) algorithms have been demonstrated to be effective for both the image reconstruction task and the motion correction task, but the two tasks are considered separately. The image reconstruction task involves removing undersampling artifacts such as noise and aliasing artifacts, whereas motion correction involves removing artifacts including blurring, ghosting, and ringing. In this work, we propose a novel method to simultaneously accelerate imaging and correct motion. This is achieved by integrating a motion module into the deep learning-based MRI reconstruction process, enabling real-time detection and correction of motion. We model motion as a tightly integrated auxiliary layer in the deep learning model during training, making the deep learning model 'motion-informed'. During inference, image reconstruction is performed from undersampled raw k-space data using a trained motion-informed DL model. Experimental results demonstrate that the proposed motion-informed deep learning image reconstruction network outperformed the conventional image reconstruction network for motion-degraded MRI datasets.
Authors: Yingwen Wu, Ruiji Yu, Xinwen Cheng, Zhengbao He, Xiaolin Huang
Abstract: In the open world, detecting out-of-distribution (OOD) data, whose labels are disjoint with those of in-distribution (ID) samples, is important for reliable deep neural networks (DNNs). To achieve better detection performance, one type of approach proposes to fine-tune the model with auxiliary OOD datasets to amplify the difference between ID and OOD data through a separation loss defined on model outputs. However, none of these studies consider enlarging the feature disparity, which should be more effective compared to outputs. The main difficulty lies in the diversity of OOD samples, which makes it hard to describe their feature distribution, let alone design losses to separate them from ID features. In this paper, we neatly fence off the problem based on an aggregation property of ID features named Neural Collapse (NC). NC means that the penultimate features of ID samples within a class are nearly identical to the last layer weight of the corresponding class. Based on this property, we propose a simple but effective loss called OrthLoss, which binds the features of OOD data in a subspace orthogonal to the principal subspace of ID features formed by NC. In this way, the features of ID and OOD samples are separated by different dimensions. By optimizing the feature separation loss rather than purely enlarging output differences, our detection achieves SOTA performance on CIFAR benchmarks without any additional data augmentation or sampling, demonstrating the importance of feature separation in OOD detection. The code will be published.
Authors: Yuka Hashimoto, Ayoub Hafid, Masahiro Ikeda, Hachem Kadri
Abstract: In this paper, we propose a new class of positive definite kernels based on the spectral truncation, which has been discussed in the fields of noncommutative geometry and $C^*$-algebra. We focus on kernels whose inputs and outputs are functions and generalize existing kernels, such as polynomial, product, and separable kernels, by introducing a truncation parameter $n$ that describes the noncommutativity of the products appearing in the kernels. When $n$ goes to infinity, the proposed kernels tend to the existing commutative kernels. If $n$ is finite, they exhibit different behavior, and the noncommutativity induces interactions along the data function domain. We show that the truncation parameter $n$ is a governing factor leading to performance enhancement: by setting an appropriate $n$, we can balance the representation power and the complexity of the representation space. The flexibility of the proposed class of kernels allows us to go beyond previous commutative kernels.
Authors: Zavareh Bozorgasl, Hao Chen
Abstract: This paper presents the development and application of Wavelet Kolmogorov-Arnold Networks (Wav-KAN) in federated learning. We implemented Wav-KAN \cite{wav-kan} in the clients. Indeed, we have considered both continuous wavelet transform (CWT) and also discrete wavelet transform (DWT) to enable multiresolution capabaility which helps in heteregeneous data distribution across clients. Extensive experiments were conducted on different datasets, demonstrating Wav-KAN's superior performance in terms of interpretability, computational speed, training and test accuracy. Our federated learning algorithm integrates wavelet-based activation functions, parameterized by weight, scale, and translation, to enhance local and global model performance. Results show significant improvements in computational efficiency, robustness, and accuracy, highlighting the effectiveness of wavelet selection in scalable neural network design.
Authors: Akio Hayakawa, Masato Ishii, Takashi Shibuya, Yuki Mitsufuji
Abstract: In this study, we aim to construct an audio-video generative model with minimal computational cost by leveraging pre-trained single-modal generative models for audio and video. To achieve this, we propose a novel method that guides each single-modal model to cooperatively generate well-aligned samples across modalities. Specifically, given two pre-trained base diffusion models, we train a lightweight joint guidance module to adjust scores separately estimated by the base models to match the score of joint distribution over audio and video. We theoretically show that this guidance can be computed through the gradient of the optimal discriminator distinguishing real audio-video pairs from fake ones independently generated by the base models. On the basis of this analysis, we construct the joint guidance module by training this discriminator. Additionally, we adopt a loss function to make the gradient of the discriminator work as a noise estimator, as in standard diffusion models, stabilizing the gradient of the discriminator. Empirical evaluations on several benchmark datasets demonstrate that our method improves both single-modal fidelity and multi-modal alignment with a relatively small number of parameters.
Authors: Yen-An Lu, Wei-Shou Hu, Joel A. Paulson, Qi Zhang
Abstract: This work addresses data-driven inverse optimization (IO), where the goal is to estimate unknown parameters in an optimization model from observed decisions that can be assumed to be optimal or near-optimal solutions to the optimization problem. The IO problem is commonly formulated as a large-scale bilevel program that is notoriously difficult to solve. Deviating from traditional exact solution methods, we propose a derivative-free optimization approach based on Bayesian optimization, which we call BO4IO, to solve general IO problems. We treat the IO loss function as a black box and approximate it with a Gaussian process model. Using the predicted posterior function, an acquisition function is minimized at each iteration to query new candidate solutions and sequentially converge to the optimal parameter estimates. The main advantages of using Bayesian optimization for IO are two-fold: (i) it circumvents the need of complex reformulations of the bilevel program or specialized algorithms and can hence enable computational tractability even when the underlying optimization problem is nonconvex or involves discrete variables, and (ii) it allows approximations of the profile likelihood, which provide uncertainty quantification on the IO parameter estimates. We apply the proposed method to three computational case studies, covering different classes of forward optimization problems ranging from convex nonlinear to nonconvex mixed-integer nonlinear programs. Our extensive computational results demonstrate the efficacy and robustness of BO4IO to accurately estimate unknown model parameters from small and noisy datasets. In addition, the proposed profile likelihood analysis has proven to be effective in providing good approximations of the confidence intervals on the parameter estimates and assessing the identifiability of the unknown parameters.
Authors: Wujiang Xu, Zujie Liang, Jiaojiao Han, Xuying Ning, Wenfang Lin, Linxun Chen, Feng Wei, Yongfeng Zhang
Abstract: The sequential Recommendation (SR) task involves predicting the next item a user is likely to interact with, given their past interactions. The SR models examine the sequence of a user's actions to discern more complex behavioral patterns and temporal dynamics. Recent research demonstrates the great impact of LLMs on sequential recommendation systems, either viewing sequential recommendation as language modeling or serving as the backbone for user representation. Although these methods deliver outstanding performance, there is scant evidence of the necessity of a large language model and how large the language model is needed, especially in the sequential recommendation scene. Meanwhile, due to the huge size of LLMs, it is inefficient and impractical to apply a LLM-based model in real-world platforms that often need to process billions of traffic logs daily. In this paper, we explore the influence of LLMs' depth by conducting extensive experiments on large-scale industry datasets. Surprisingly, we discover that most intermediate layers of LLMs are redundant. Motivated by this insight, we empower small language models for SR, namely SLMRec, which adopt a simple yet effective knowledge distillation method. Moreover, SLMRec is orthogonal to other post-training efficiency techniques, such as quantization and pruning, so that they can be leveraged in combination. Comprehensive experimental results illustrate that the proposed SLMRec model attains the best performance using only 13% of the parameters found in LLM-based recommendation models, while simultaneously achieving up to 6.6x and 8.0x speedups in training and inference time costs, respectively.
Authors: Yaoyao Xu, Xinjian Zhao, Xiaozhuang Song, Benyou Wang, Tianshu Yu
Abstract: We introduce a pioneering methodology for boosting large language models in the domain of protein representation learning. Our primary contribution lies in the refinement process for correlating the over-reliance on co-evolution knowledge, in a way that networks are trained to distill invaluable insights from negative samples, constituted by protein pairs sourced from disparate categories. By capitalizing on this novel approach, our technique steers the training of transformer-based models within the attention score space. This advanced strategy not only amplifies performance but also reflects the nuanced biological behaviors exhibited by proteins, offering aligned evidence with traditional biological mechanisms such as protein-protein interaction. We experimentally observed improved performance on various tasks over datasets, on top of several well-established large protein models. This innovative paradigm opens up promising horizons for further progress in the realms of protein research and computational biology.
Authors: Zhengji Li, Xi Xiao, Jiacheng Xie, Yuxiao Fan, Wentao Wang, Gang Chen, Liqiang Zhang, Tianyang Wang
Abstract: With the development of modern society, traffic volume continues to increase in most countries worldwide, leading to an increase in the rate of pavement damage Therefore, the real-time and highly accurate pavement damage detection and maintenance have become the current need. In this paper, an enhanced pavement damage detection method with CycleGAN and improved YOLOv5 algorithm is presented. We selected 7644 self-collected images of pavement damage samples as the initial dataset and augmented it by CycleGAN. Due to a substantial difference between the images generated by CycleGAN and real road images, we proposed a data enhancement method based on an improved Scharr filter, CycleGAN, and Laplacian pyramid. To improve the target recognition effect on a complex background and solve the problem that the spatial pyramid pooling-fast module in the YOLOv5 network cannot handle multiscale targets, we introduced the convolutional block attention module attention mechanism and proposed the atrous spatial pyramid pooling with squeeze-and-excitation structure. In addition, we optimized the loss function of YOLOv5 by replacing the CIoU with EIoU. The experimental results showed that our algorithm achieved a precision of 0.872, recall of 0.854, and mean average precision@0.5 of 0.882 in detecting three main types of pavement damage: cracks, potholes, and patching. On the GPU, its frames per second reached 68, meeting the requirements for real-time detection. Its overall performance even exceeded the current more advanced YOLOv7 and achieved good results in practical applications, providing a basis for decision-making in pavement damage detection and prevention.
Authors: Shakti N. Wadekar, Abhishek Chaurasia, Aman Chadha, Eugenio Culurciello
Abstract: This work uniquely identifies and characterizes four prevalent multimodal model architectural patterns in the contemporary multimodal landscape. Systematically categorizing models by architecture type facilitates monitoring of developments in the multimodal domain. Distinct from recent survey papers that present general information on multimodal architectures, this research conducts a comprehensive exploration of architectural details and identifies four specific architectural types. The types are distinguished by their respective methodologies for integrating multimodal inputs into the deep neural network model. The first two types (Type A and B) deeply fuses multimodal inputs within the internal layers of the model, whereas the following two types (Type C and D) facilitate early fusion at the input stage. Type-A employs standard cross-attention, whereas Type-B utilizes custom-designed layers for modality fusion within the internal layers. On the other hand, Type-C utilizes modality-specific encoders, while Type-D leverages tokenizers to process the modalities at the model's input stage. The identified architecture types aid the monitoring of any-to-any multimodal model development. Notably, Type-C and Type-D are currently favored in the construction of any-to-any multimodal models. Type-C, distinguished by its non-tokenizing multimodal model architecture, is emerging as a viable alternative to Type-D, which utilizes input-tokenizing techniques. To assist in model selection, this work highlights the advantages and disadvantages of each architecture type based on data and compute requirements, architecture complexity, scalability, simplification of adding modalities, training objectives, and any-to-any multimodal generation capability.
Authors: Keming Lu, Bowen Yu, Fei Huang, Yang Fan, Runji Lin, Chang Zhou
Abstract: Effectively aligning Large Language Models (LLMs) with human-centric values while preventing the degradation of abilities acquired through Pre-training and Supervised Fine-tuning (SFT) poses a central challenge in Reinforcement Learning from Human Feedback (RLHF). In this paper, we first discover that interpolating RLHF and SFT model parameters can adjust the trade-off between human preference and basic capabilities, thereby reducing the alignment tax at the cost of alignment reward. Inspired by this, we propose integrating the RL policy and SFT models at each optimization step in RLHF to continuously regulate the training direction, introducing the Online Merging Optimizer. Specifically, we merge gradients with the parameter differences between SFT and pretrained models, effectively steering the gradient towards maximizing rewards in the direction of SFT optimization. We demonstrate that our optimizer works well with different LLM families, such as Qwen and LLaMA, across various model sizes ranging from 1.8B to 8B, various RLHF algorithms like DPO and KTO, and existing model merging methods. It significantly enhances alignment reward while mitigating alignment tax, achieving higher overall performance across 14 benchmarks.
Authors: O. Deniz Akyildiz, Mark Girolami, Andrew M. Stuart, Arnaud Vadeboncoeur
Abstract: Bayesian inversion is central to the quantification of uncertainty within problems arising from numerous applications in science and engineering. To formulate the approach, four ingredients are required: a forward model mapping the unknown parameter to an element of a solution space, often the solution space for a differential equation; an observation operator mapping an element of the solution space to the data space; a noise model describing how noise pollutes the observations; and a prior model describing knowledge about the unknown parameter before the data is acquired. This paper is concerned with learning the prior model from data; in particular, learning the prior from multiple realizations of indirect data obtained through the noisy observation process. The prior is represented, using a generative model, as the pushforward of a Gaussian in a latent space; the pushforward map is learned by minimizing an appropriate loss function. A metric that is well-defined under empirical approximation is used to define the loss function for the pushforward map to make an implementable methodology. Furthermore, an efficient residual-based neural operator approximation of the forward model is proposed and it is shown that this may be learned concurrently with the pushforward map, using a bilevel optimization formulation of the problem; this use of neural operator approximation has the potential to make prior learning from indirect data more computationally efficient, especially when the observation process is expensive, non-smooth or not known. The ideas are illustrated with the Darcy flow inverse problem of finding permeability from piezometric head measurements.
Authors: Yunzhi Yao, Ningyu Zhang, Zekun Xi, Mengru Wang, Ziwen Xu, Shumin Deng, Huajun Chen
Abstract: The remarkable capabilities of modern large language models are rooted in their vast repositories of knowledge encoded within their parameters, enabling them to perceive the world and engage in reasoning. The inner workings of how these models store knowledge have long been a subject of intense interest and investigation among researchers. To date, most studies have concentrated on isolated components within these models, such as the Multilayer Perceptrons and attention head. In this paper, we delve into the computation graph of the language model to uncover the knowledge circuits that are instrumental in articulating specific knowledge. The experiments, conducted with GPT2 and TinyLLAMA, has allowed us to observe how certain information heads, relation heads, and Multilayer Perceptrons collaboratively encode knowledge within the model. Moreover, we evaluate the impact of current knowledge editing techniques on these knowledge circuits, providing deeper insights into the functioning and constraints of these editing methodologies. Finally, we utilize knowledge circuits to analyze and interpret language model behaviors such as hallucinations and in-context learning. We believe the knowledge circuit holds potential for advancing our understanding of Transformers and guiding the improved design of knowledge editing. Code and data are available in https://github.com/zjunlp/KnowledgeCircuits.
Authors: Shentong Mo, Sukmin Yun
Abstract: The joint-embedding predictive architecture (JEPA) recently has shown impressive results in extracting visual representations from unlabeled imagery under a masking strategy. However, we reveal its disadvantages, notably its insufficient understanding of local semantics. This deficiency originates from masked modeling in the embedding space, resulting in a reduction of discriminative power and can even lead to the neglect of critical local semantics. To bridge this gap, we introduce DMT-JEPA, a novel masked modeling objective rooted in JEPA, specifically designed to generate discriminative latent targets from neighboring information. Our key idea is simple: we consider a set of semantically similar neighboring patches as a target of a masked patch. To be specific, the proposed DMT-JEPA (a) computes feature similarities between each masked patch and its corresponding neighboring patches to select patches having semantically meaningful relations, and (b) employs lightweight cross-attention heads to aggregate features of neighboring patches as the masked targets. Consequently, DMT-JEPA demonstrates strong discriminative power, offering benefits across a diverse spectrum of downstream tasks. Through extensive experiments, we demonstrate our effectiveness across various visual benchmarks, including ImageNet-1K image classification, ADE20K semantic segmentation, and COCO object detection tasks. Code is available at: \url{https://github.com/DMTJEPA/DMTJEPA}.
Authors: Zican Dong, Junyi Li, Xin Men, Wayne Xin Zhao, Bingbing Wang, Zhen Tian, Weipeng Chen, Ji-Rong Wen
Abstract: Transformer-based large language models (LLMs) typically have a limited context window, resulting in significant performance degradation when processing text beyond the length of the context window. Extensive studies have been proposed to extend the context window and achieve length extrapolation of LLMs, but there is still a lack of in-depth interpretation of these approaches. In this study, we explore the positional information within and beyond the context window for deciphering the underlying mechanism of LLMs. By using a mean-based decomposition method, we disentangle positional vectors from hidden states of LLMs and analyze their formation and effect on attention. Furthermore, when texts exceed the context window, we analyze the change of positional vectors in two settings, i.e., direct extrapolation and context window extension. Based on our findings, we design two training-free context window extension methods, positional vector replacement and attention window extension. Experimental results show that our methods can effectively extend the context window length.
Authors: Zebin You, Xinyu Zhang, Hanzhong Guo, Jingdong Wang, Chongxuan Li
Abstract: The ultimate goal of generative models is to characterize the data distribution perfectly. For image generation, common metrics of visual quality (e.g., FID), and the truthlikeness of generated images to the human eyes seem to suggest that we are close to achieving it. However, through distribution classification tasks, we find that, in the eyes of classifiers parameterized by neural networks, the strongest diffusion models are still far from this goal. Specifically, classifiers consistently and effortlessly distinguish between real and generated images in various settings. Further, we observe an intriguing discrepancy: classifiers can identify differences between diffusion models with similar performance (e.g., U-ViT-H vs. DiT-XL), but struggle to differentiate between the smallest and largest models in the same family (e.g., EDM2-XS vs. EDM2-XXL), whereas humans exhibit the opposite tendency. As an explanation, our comprehensive empirical study suggests that, unlike humans, classifiers tend to classify images through edge and high-frequency components. We believe that our methodology can serve as a probe to understand how generative models work and inspire further thought on how existing models can be improved and how the abuse of such models can be prevented.
Authors: Dmitry Kovalev, Ekaterina Borodich, Alexander Gasnikov, Dmitrii Feoktistov
Abstract: We consider the task of minimizing the sum of convex functions stored in a decentralized manner across the nodes of a communication network. This problem is relatively well-studied in the scenario when the objective functions are smooth, or the links of the network are fixed in time, or both. In particular, lower bounds on the number of decentralized communications and (sub)gradient computations required to solve the problem have been established, along with matching optimal algorithms. However, the remaining and most challenging setting of non-smooth decentralized optimization over time-varying networks is largely underexplored, as neither lower bounds nor optimal algorithms are known in the literature. We resolve this fundamental gap with the following contributions: (i) we establish the first lower bounds on the communication and subgradient computation complexities of solving non-smooth convex decentralized optimization problems over time-varying networks; (ii) we develop the first optimal algorithm that matches these lower bounds and offers substantially improved theoretical performance compared to the existing state of the art.
Authors: Youngwan Lee, Jeffrey Ryan Willette, Jonghee Kim, Sung Ju Hwang
Abstract: The Masked autoencoder (MAE) has drawn attention as a representative self-supervised approach for masked image modeling with vision transformers. However, even though MAE shows better generalization capability than fully supervised training from scratch, the reason why has not been explored. In another line of work, the Reconstruction Consistent Masked Auto Encoder (RC-MAE), has been proposed which adopts a self-distillation scheme in the form of an exponential moving average (EMA) teacher into MAE, and it has been shown that the EMA-teacher performs a conditional gradient correction during optimization. To further investigate the reason for better generalization of the self-supervised ViT when trained by MAE (MAE-ViT) and the effect of the gradient correction of RC-MAE from the perspective of optimization, we visualize the loss landscapes of the self-supervised vision transformer by both MAE and RC-MAE and compare them with the supervised ViT (Sup-ViT). Unlike previous loss landscape visualizations of neural networks based on classification task loss, we visualize the loss landscape of ViT by computing pre-training task loss. Through the lens of loss landscapes, we find two interesting observations: (1) MAE-ViT has a smoother and wider overall loss curvature than Sup-ViT. (2) The EMA-teacher allows MAE to widen the region of convexity in both pretraining and linear probing, leading to quicker convergence. To the best of our knowledge, this work is the first to investigate the self-supervised ViT through the lens of the loss landscape.
Authors: Hind I. Alshbanat, Hafida Benhidour, Said Kerrache
Abstract: Recommender systems are essential tools in the digital era, providing personalized content to users in areas like e-commerce, entertainment, and social media. Among the many approaches developed to create these systems, latent factor models have proven particularly effective. This survey systematically reviews latent factor models in recommender systems, focusing on their core principles, methodologies, and recent advancements. The literature is examined through a structured framework covering learning data, model architecture, learning strategies, and optimization techniques. The analysis includes a taxonomy of contributions and detailed discussions on the types of learning data used, such as implicit feedback, trust, and content data, various models such as probabilistic, nonlinear, and neural models, and an exploration of diverse learning strategies like online learning, transfer learning, and active learning. Furthermore, the survey addresses the optimization strategies used to train latent factor models, improving their performance and scalability. By identifying trends, gaps, and potential research directions, this survey aims to provide valuable insights for researchers and practitioners looking to advance the field of recommender systems.
Authors: Henry W J Reeve
Abstract: We consider a semi-supervised classification problem with non-stationary label-shift in which we observe a labelled data set followed by a sequence of unlabelled covariate vectors in which the marginal probabilities of the class labels may change over time. Our objective is to predict the corresponding class-label for each covariate vector, without ever observing the ground-truth labels, beyond the initial labelled data set. Previous work has demonstrated the potential of sophisticated variants of online gradient descent to perform competitively with the optimal dynamic strategy (Bai et al. 2022). In this work we explore an alternative approach grounded in statistical methods for adaptive transfer learning. We demonstrate the merits of this alternative methodology by establishing a high-probability regret bound on the test error at any given individual test-time, which adapt automatically to the unknown dynamics of the marginal label probabilities. Further more, we give bounds on the average dynamic regret which match the average guarantees of the online learning perspective for any given time interval.
Authors: Jinkyu Yim, Jaeyong Song, Yerim Choi, Jaebeen Lee, Jaewon Jung, Hongsun Jang, Jinho Lee
Abstract: Training large language models (LLMs) is known to be challenging because of the huge computational and memory capacity requirements. To address these issues, it is common to use a cluster of GPUs with 3D parallelism, which splits a model along the data batch, pipeline stage, and intra-layer tensor dimensions. However, the use of 3D parallelism produces the additional challenge of finding the optimal number of ways on each dimension and mapping the split models onto the GPUs. Several previous studies have attempted to automatically find the optimal configuration, but many of these lacked several important aspects. For instance, the heterogeneous nature of the interconnect speeds is often ignored. While the peak bandwidths for the interconnects are usually made equal, the actual attained bandwidth varies per link in real-world clusters. Combined with the critical path modeling that does not properly consider the communication, they easily fall into sub-optimal configurations. In addition, they often fail to consider the memory requirement per GPU, often recommending solutions that could not be executed. To address these challenges, we propose Pipette, which is an automatic fine-grained LLM training configurator for real-world clusters. By devising better performance models along with the memory estimator and fine-grained individual GPU assignment, Pipette achieves faster configurations that satisfy the memory constraints. We evaluated Pipette on large clusters to show that it provides a significant speedup over the prior art. The implementation of Pipette is available at https://github.com/yimjinkyu1/date2024_pipette.
Authors: David W. Hogg (NYU, MPIA, Flatiron), Soledad Villar (JHU, Flatiron)
Abstract: Machine learning (ML) methods are having a huge impact across all of the sciences. However, ML has a strong ontology - in which only the data exist - and a strong epistemology - in which a model is considered good if it performs well on held-out training data. These philosophies are in strong conflict with both standard practices and key philosophies in the natural sciences. Here, we identify some locations for ML in the natural sciences at which the ontology and epistemology are valuable. For example, when an expressive machine learning model is used in a causal inference to represent the effects of confounders, such as foregrounds, backgrounds, or instrument calibration parameters, the model capacity and loose philosophy of ML can make the results more trustworthy. We also show that there are contexts in which the introduction of ML introduces strong, unwanted statistical biases. For one, when ML models are used to emulate physical (or first-principles) simulations, they introduce strong confirmation biases. For another, when expressive regressions are used to label datasets, those labels cannot be used in downstream joint or ensemble analyses without taking on uncontrolled biases. The question in the title is being asked of all of the natural sciences; that is, we are calling on the scientific communities to take a step back and consider the role and value of ML in their fields; the (partial) answers we give here come from the particular perspective of physics.
Authors: Wei Cheng, Hongrui Ye, Xiao Wen, Jiachen Zhang, Jiping Xu, Feifan Zhang
Abstract: Deep learning has significantly improved the accuracy of crop classification using multispectral temporal data. However, these models have complex structures with numerous parameters, requiring large amounts of data and costly training. In low-resource situations with fewer labeled samples, deep learning models perform poorly due to insufficient data. Conversely, compressors are data-type agnostic, and non-parametric methods do not bring underlying assumptions. Inspired by this insight, we propose a non-training alternative to deep learning models, aiming to address these situations. Specifically, the Symbolic Representation Module is proposed to convert the reflectivity into symbolic representations. The symbolic representations are then cross-transformed in both the channel and time dimensions to generate symbolic embeddings. Next, the Multi-scale Normalised Compression Distance (MNCD) is designed to measure the correlation between any two symbolic embeddings. Finally, based on the MNCDs, high quality crop classification can be achieved using only a k-nearest-neighbor classifier kNN. The entire framework is ready-to-use and lightweight. Without any training, it outperformed, on average, 7 advanced deep learning models trained at scale on three benchmark datasets. It also outperforms more than half of these models in the few-shot setting with sparse crop labels. Therefore, the high performance and robustness of our non-training framework makes it truly applicable to real-world crop mapping. Codes are available at: https://github.com/qinfengsama/Compressor-Based-Crop-Mapping.
URLs: https://github.com/qinfengsama/Compressor-Based-Crop-Mapping.
Authors: Hao Yu, Minghao Fu, Jiandong Ding, Yusheng Zhou, Jianxin Wu
Abstract: Deep Click-Through Rate (CTR) prediction models play an important role in modern industrial recommendation scenarios. However, high memory overhead and computational costs limit their deployment in resource-constrained environments. Low-rank approximation is an effective method for computer vision and natural language processing models, but its application in compressing CTR prediction models has been less explored. Due to the limited memory and computing resources, compression of CTR prediction models often confronts three fundamental challenges, i.e., (1). How to reduce the model sizes to adapt to edge devices? (2). How to speed up CTR prediction model inference? (3). How to retain the capabilities of original models after compression? Previous low-rank compression research mostly uses tensor decomposition, which can achieve a high parameter compression ratio, but brings in AUC degradation and additional computing overhead. To address these challenges, we propose a unified low-rank decomposition framework for compressing CTR prediction models. We find that even with the most classic matrix decomposition SVD method, our framework can achieve better performance than the original model. To further improve the effectiveness of our framework, we locally compress the output features instead of compressing the model weights. Our unified low-rank compression framework can be applied to embedding tables and MLP layers in various CTR prediction models. Extensive experiments on two academic datasets and one real industrial benchmark demonstrate that, with 3-5x model size reduction, our compressed models can achieve both faster inference and higher AUC than the uncompressed original models. Our code is at https://github.com/yuhao318/Atomic_Feature_Mimicking.
Authors: Javier Naranjo-Alcazar, Jordi Grau-Haro, Ruben Ribes-Serrano, Pedro Zuccarello
Abstract: Machine Listening focuses on developing technologies to extract relevant information from audio signals. A critical aspect of these projects is the acquisition and labeling of contextualized data, which is inherently complex and requires specific resources and strategies. Despite the availability of some audio datasets, many are unsuitable for commercial applications. The paper emphasizes the importance of Active Learning (AL) using expert labelers over crowdsourcing, which often lacks detailed insights into dataset structures. AL is an iterative process combining human labelers and AI models to optimize the labeling budget by intelligently selecting samples for human review. This approach addresses the challenge of handling large, constantly growing datasets that exceed available computational resources and memory. The paper presents a comprehensive data-centric framework for Machine Listening projects, detailing the configuration of recording nodes, database structure, and labeling budget optimization in resource-constrained scenarios. Applied to an industrial port in Valencia, Spain, the framework successfully labeled 6540 ten-second audio samples over five months with a small team, demonstrating its effectiveness and adaptability to various resource availability situations.
Authors: Yuhan Li, Hao Zhou, Wenxiang Shang, Ran Lin, Xuanhong Chen, Bingbing Ni
Abstract: While image-based virtual try-on has made significant strides, emerging approaches still fall short of delivering high-fidelity and robust fitting images across various scenarios, as their models suffer from issues of ill-fitted garment styles and quality degrading during the training process, not to mention the lack of support for various combinations of attire. Therefore, we first propose a lightweight, scalable, operator known as Hydra Block for attire combinations. This is achieved through a parallel attention mechanism that facilitates the feature injection of multiple garments from conditionally encoded branches into the main network. Secondly, to significantly enhance the model's robustness and expressiveness in real-world scenarios, we evolve its potential across diverse settings by synthesizing the residuals of multiple models, as well as implementing a mask region boost strategy to overcome the instability caused by information leakage in existing models. Equipped with the above design, AnyFit surpasses all baselines on high-resolution benchmarks and real-world data by a large gap, excelling in producing well-fitting garments replete with photorealistic and rich details. Furthermore, AnyFit's impressive performance on high-fidelity virtual try-ons in any scenario from any image, paves a new path for future research within the fashion community.
Authors: Ilia Azizi, Marc-Olivier Boldi, Val\'erie Chavez-Demoulin
Abstract: This work introduces the Supervised Expectation-Maximization Framework (SEMF), a versatile and model-agnostic framework that generates prediction intervals for datasets with complete or missing data. SEMF extends the Expectation-Maximization (EM) algorithm, traditionally used in unsupervised learning, to a supervised context, enabling it to extract latent representations for uncertainty estimation. The framework demonstrates robustness through extensive empirical evaluation across 11 tabular datasets, achieving$\unicode{x2013}$in some cases$\unicode{x2013}$narrower normalized prediction intervals and higher coverage than traditional quantile regression methods. Furthermore, SEMF integrates seamlessly with existing machine learning algorithms, such as gradient-boosted trees and neural networks, exemplifying its usefulness for real-world applications. The experimental results highlight SEMF's potential to advance state-of-the-art techniques in uncertainty quantification.
Authors: Daniel Bethell, Simos Gerasimou, Radu Calinescu, Calum Imrie
Abstract: Empowering safe exploration of reinforcement learning (RL) agents during training is a critical impediment towards deploying RL agents in many real-world scenarios. Training RL agents in unknown, black-box environments poses an even greater safety risk when prior knowledge of the domain/task is unavailable. We introduce ADVICE (Adaptive Shielding with a Contrastive Autoencoder), a novel post-shielding technique that distinguishes safe and unsafe features of state-action pairs during training, thus protecting the RL agent from executing actions that yield potentially hazardous outcomes. Our comprehensive experimental evaluation against state-of-the-art safe RL exploration techniques demonstrates how ADVICE can significantly reduce safety violations during training while maintaining a competitive outcome reward.
Authors: Vitalis Vosylius, Younggyo Seo, Jafar Uru\c{c}, Stephen James
Abstract: In the field of Robot Learning, the complex mapping between high-dimensional observations such as RGB images and low-level robotic actions, two inherently very different spaces, constitutes a complex learning problem, especially with limited amounts of data. In this work, we introduce Render and Diffuse (R&D) a method that unifies low-level robot actions and RGB observations within the image space using virtual renders of the 3D model of the robot. Using this joint observation-action representation it computes low-level robot actions using a learnt diffusion process that iteratively updates the virtual renders of the robot. This space unification simplifies the learning problem and introduces inductive biases that are crucial for sample efficiency and spatial generalisation. We thoroughly evaluate several variants of R&D in simulation and showcase their applicability on six everyday tasks in the real world. Our results show that R&D exhibits strong spatial generalisation capabilities and is more sample efficient than more common image-to-action methods.
Authors: Chengxing Xie, Difan Zou
Abstract: Recent studies have highlighted their proficiency in some simple tasks like writing and coding through various reasoning strategies. However, LLM agents still struggle with tasks that require comprehensive planning, a process that challenges current models and remains a critical research issue. In this study, we concentrate on travel planning, a Multi-Phases planning problem, that involves multiple interconnected stages, such as outlining, information gathering, and planning, often characterized by the need to manage various constraints and uncertainties. Existing reasoning approaches have struggled to effectively address this complex task. Our research aims to address this challenge by developing a human-like planning framework for LLM agents, i.e., guiding the LLM agent to simulate various steps that humans take when solving Multi-Phases problems. Specifically, we implement several strategies to enable LLM agents to generate a coherent outline for each travel query, mirroring human planning patterns. Additionally, we integrate Strategy Block and Knowledge Block into our framework: Strategy Block facilitates information collection, while Knowledge Block provides essential information for detailed planning. Through our extensive experiments, we demonstrate that our framework significantly improves the planning capabilities of LLM agents, enabling them to tackle the travel planning task with improved efficiency and effectiveness. Our experimental results showcase the exceptional performance of the proposed framework; when combined with GPT-4-Turbo, it attains $10\times$ the performance gains in comparison to the baseline framework deployed on GPT-4-Turbo.
Authors: Zhi Zheng, Shangding Gu
Abstract: Ensuring safety in MARL, particularly when deploying it in real-world applications such as autonomous driving, emerges as a critical challenge. To address this challenge, traditional safe MARL methods extend MARL approaches to incorporate safety considerations, aiming to minimize safety risk values. However, these safe MARL algorithms often fail to model other agents and lack convergence guarantees, particularly in dynamically complex environments. In this study, we propose a safe MARL method grounded in a Stackelberg model with bi-level optimization, for which convergence analysis is provided. Derived from our theoretical analysis, we develop two practical algorithms, namely Constrained Stackelberg Q-learning (CSQ) and Constrained Stackelberg Multi-Agent Deep Deterministic Policy Gradient (CS-MADDPG), designed to facilitate MARL decision-making in autonomous driving applications. To evaluate the effectiveness of our algorithms, we developed a safe MARL autonomous driving benchmark and conducted experiments on challenging autonomous driving scenarios, such as merges, roundabouts, intersections, and racetracks. The experimental results indicate that our algorithms, CSQ and CS-MADDPG, outperform several strong MARL baselines, such as Bi-AC, MACPO, and MAPPO-L, regarding reward and safety performance. The demos and source code are available at {https://github.com/SafeRL-Lab/Safe-MARL-in-Autonomous-Driving.git}.
URLs: https://github.com/SafeRL-Lab/Safe-MARL-in-Autonomous-Driving.git
Authors: Kazu Ghalamkari, Jesper L{\o}ve Hinrich, Morten M{\o}rup
Abstract: We present an expectation-maximization (EM) based unified framework for non-negative tensor decomposition that optimizes the Kullback-Leibler divergence. To avoid iterations in each M-step and learning rate tuning, we establish a general relationship between low-rank decomposition and many-body approximation. Using this connection, we exploit that the closed-form solution of the many-body approximation can be used to update all parameters simultaneously in the M-step. Our framework not only offers a unified methodology for a variety of low-rank structures, including CP, Tucker, and Train decompositions, but also their combinations forming mixtures of tensors as well as robust adaptive noise modeling. Empirically, we demonstrate that our framework provides superior generalization for discrete density estimation compared to conventional tensor-based approaches.
Authors: Semih Cayci, Atilla Eryilmaz
Abstract: In this paper, we study a natural policy gradient method based on recurrent neural networks (RNNs) for partially-observable Markov decision processes, whereby RNNs are used for policy parameterization and policy evaluation to address curse of dimensionality in non-Markovian reinforcement learning. We present finite-time and finite-width analyses for both the critic (recurrent temporal difference learning), and correspondingly-operated recurrent natural policy gradient method in the near-initialization regime. Our analysis demonstrates the efficiency of RNNs for problems with short-term memory with explicit bounds on the required network widths and sample complexity, and points out the challenges in the case of long-term dependencies.
Authors: Ivan Petrukha, Nataliia Stulova, Sergii Kryvoblotskyi
Abstract: Background. The recent surge in phishing attacks keeps undermining the effectiveness of the traditional anti-phishing blacklist approaches. On-device anti-phishing solutions are gaining popularity as they offer faster phishing detection locally. Aim. We aim to eliminate the delay in recognizing and recording phishing campaigns in databases via on-device solutions that identify phishing sites immediately when encountered by the user rather than waiting for a web crawler's scan to finish. Additionally, utilizing operating system-specific resources and frameworks, we aim to minimize the impact on system performance and depend on local processing to protect user privacy. Method. We propose a phishing detection solution that uses a combination of computer vision and on-device machine learning models to analyze websites in real time. Our reference-based approach analyzes the visual content of webpages, identifying phishing attempts through layout analysis, credential input areas detection, and brand impersonation criteria combination. Results. Our case study shows it's feasible to perform background processing on-device continuously, for the case of the web browser requiring the resource use of 16% of a single CPU core and less than 84MB of RAM on Apple M1 while maintaining the accuracy of brand logo detection at 46.6% (comparable with baselines), and of Credential Requiring Page detection at 98.1% (improving the baseline by 3.1%), within the test dataset. Conclusions. Our results demonstrate the potential of on-device, real-time phishing detection systems to enhance cybersecurity defensive technologies and extend the scope of phishing detection to more similar regions of interest, e.g., email clients and messenger windows.
Authors: Reihaneh Teimouri, Marta Kersten-Oertel, Yiming Xiao
Abstract: Efficient and accurate brain ventricle segmentation from clinical CT scans is critical for emergency surgeries like ventriculostomy. With the challenges in poor soft tissue contrast and a scarcity of well-annotated databases for clinical brain CTs, we introduce a novel uncertainty-aware ventricle segmentation technique without the need of CT segmentation ground truths by leveraging diffusion-model-based domain adaptation. Specifically, our method employs the diffusion Schr\"odinger Bridge and an attention recurrent residual U-Net to capitalize on unpaired CT and MRI scans to derive automatic CT segmentation from those of the MRIs, which are more accessible. Importantly, we propose an end-to-end, joint training framework of image translation and segmentation tasks, and demonstrate its benefit over training individual tasks separately. By comparing the proposed method against similar setups using two different GAN models for domain adaptation (CycleGAN and CUT), we also reveal the advantage of diffusion models towards improved segmentation and image translation quality. With a Dice score of 0.78$\pm$0.27, our proposed method outperformed the compared methods, including SynSeg-Net, while providing intuitive uncertainty measures to further facilitate quality control of the automatic segmentation outcomes.
Authors: Christopher Criscitiello, Quentin Rebjock, Andrew D. McRae, Nicolas Boumal
Abstract: We consider the dynamics of $n$ points on a sphere in $\mathbb{R}^d$ ($d \geq 2$) which attract each other according to a function $\varphi$ of their inner products. When $\varphi$ is linear ($\varphi(t) = t$), the points converge to a common value (i.e., synchronize) in various connectivity scenarios: this is part of classical work on Kuramoto oscillator networks. When $\varphi$ is exponential ($\varphi(t) = e^{\beta t}$), these dynamics correspond to a limit of how idealized transformers process data, as described by Geshkovski et al. (2024). Accordingly, they ask whether synchronization occurs for exponential $\varphi$. In the context of consensus for multi-agent control, Markdahl et al. (2018) show that for $d \geq 3$ (spheres), if the interaction graph is connected and $\varphi$ is increasing and convex, then the system synchronizes. What is the situation on circles ($d=2$)? First, we show that $\varphi$ being increasing and convex is no longer sufficient. Then we identify a new condition (that the Taylor coefficients of $\varphi'$ are decreasing) under which we do have synchronization on the circle. In so doing, we provide some answers to the open problems posed by Geshkovski et al. (2024).
Authors: Behrad Moniri, Hamed Hassani
Abstract: In this paper, we study a nonlinear spiked random matrix model where a nonlinear function is applied element-wise to a noise matrix perturbed by a rank-one signal. We establish a signal-plus-noise decomposition for this model and identify precise phase transitions in the structure of the signal components at critical thresholds of signal strength. To demonstrate the applicability of this decomposition, we then utilize it to study new phenomena in the problems of signed signal recovery in nonlinear models and community detection in transformed stochastic block models. Finally, we validate our results through a series of numerical simulations.
Authors: Valentina Tardugno Poleo (NYU), Nora Eisner (CCA), David W. Hogg (NYU, CCA)
Abstract: Differentiating between real transit events and false positive signals in photometric time series data is a bottleneck in the identification of transiting exoplanets, particularly long-period planets. This differentiation typically requires visual inspection of a large number of transit-like signals to rule out instrumental and astrophysical false positives that mimic planetary transit signals. We build a one-dimensional convolutional neural network (CNN) to separate eclipsing binaries and other false positives from potential planet candidates, reducing the number of light curves that require human vetting. Our CNN is trained using the TESS light curves that were identified by Planet Hunters citizen scientists as likely containing a transit. We also include the background flux and centroid information. The light curves are visually inspected and labeled by project scientists and are minimally pre-processed, with only normalization and data augmentation taking place before training. The median percentage of contaminants flagged across the test sectors is 18% with a maximum of 37% and a minimum of 10%. Our model keeps 100% of the planets for 16 of the 18 test sectors, while incorrectly flagging one planet candidate (0.3%) for one sector and two (0.6%) for the remaining sector. Our method shows potential to reduce the number of light curves requiring manual vetting by up to a third with minimal misclassification of planet candidates.
Authors: Ruijian Han, Lan Luo, Yuanhang Luo, Yuanyuan Lin, Jian Huang
Abstract: Online statistical inference facilitates real-time analysis of sequentially collected data, making it different from traditional methods that rely on static datasets. This paper introduces a novel approach to online inference in high-dimensional generalized linear models, where we update regression coefficient estimates and their standard errors upon each new data arrival. In contrast to existing methods that either require full dataset access or large-dimensional summary statistics storage, our method operates in a single-pass mode, significantly reducing both time and space complexity. The core of our methodological innovation lies in an adaptive stochastic gradient descent algorithm tailored for dynamic objective functions, coupled with a novel online debiasing procedure. This allows us to maintain low-dimensional summary statistics while effectively controlling optimization errors introduced by the dynamically changing loss functions. We demonstrate that our method, termed the Approximated Debiased Lasso (ADL), not only mitigates the need for the bounded individual probability condition but also significantly improves numerical performance. Numerical experiments demonstrate that the proposed ADL method consistently exhibits robust performance across various covariance matrix structures.
Authors: Manuele Leonelli, Gherardo Varando
Abstract: Supervised classification is one of the most ubiquitous tasks in machine learning. Generative classifiers based on Bayesian networks are often used because of their interpretability and competitive accuracy. The widely used naive and TAN classifiers are specific instances of Bayesian network classifiers with a constrained underlying graph. This paper introduces novel classes of generative classifiers extending TAN and other famous types of Bayesian network classifiers. Our approach is based on staged tree models, which extend Bayesian networks by allowing for complex, context-specific patterns of dependence. We formally study the relationship between our novel classes of classifiers and Bayesian networks. We introduce and implement data-driven learning routines for our models and investigate their accuracy in an extensive computational study. The study demonstrates that models embedding asymmetric information can enhance classification accuracy.
Authors: Jaskaran Singh Walia, Pavithra L K
Abstract: Addressing the issue of submerged underwater trash is crucial for safeguarding aquatic ecosystems and preserving marine life. While identifying debris present on the surface of water bodies is straightforward, assessing the underwater submerged waste is a challenge due to the image distortions caused by factors such as light refraction, absorption, suspended particles, color shifts, and occlusion. This paper conducts a comprehensive review of state-of-the-art architectures and on the existing datasets to establish a baseline for submerged waste and trash detection. The primary goal remains to establish the benchmark of the object localization techniques to be leveraged by advanced underwater sensors and autonomous underwater vehicles. The ultimate objective is to explore the underwater environment, to identify, and remove underwater debris. The absence of benchmarks (dataset or algorithm) in many researches emphasizes the need for a more robust algorithmic solution. Through this research, we aim to give performance comparative analysis of various underwater trash detection algorithms.
Authors: Jack Storror Carter, Manuele Leonelli, Eva Riccomagno, Gherardo Varando
Abstract: Staged trees are probabilistic graphical models capable of representing any class of non-symmetric independence via a coloring of its vertices. Several structural learning routines have been defined and implemented to learn staged trees from data, under the frequentist or Bayesian paradigm. They assume a data set has been observed fully and, in practice, observations with missing entries are either dropped or imputed before learning the model. Here, we introduce the first algorithms for staged trees that handle missingness within the learning of the model. To this end, we characterize the likelihood of staged tree models in the presence of missing data and discuss pseudo-likelihoods that approximate it. A structural expectation-maximization algorithm estimating the model directly from the full likelihood is also implemented and evaluated. A computational experiment showcases the performance of the novel learning algorithms, demonstrating that it is feasible to account for different missingness patterns when learning staged trees.
Authors: Jay Jasti, Hua Zhong, Vandana Panwar, Vipul Jarmale, Jeffrey Miyata, Deyssy Carrillo, Alana Christie, Dinesh Rakheja, Zora Modrusan, Edward Ernest Kadel III, Niha Beig, Mahrukh Huseni, James Brugarolas, Payal Kapur, Satwik Rajaram
Abstract: Predictive biomarkers of treatment response are lacking for metastatic clear cell renal cell carcinoma (ccRCC), a tumor type that is treated with angiogenesis inhibitors, immune checkpoint inhibitors, mTOR inhibitors and a HIF2 inhibitor. The Angioscore, an RNA-based quantification of angiogenesis, is arguably the best candidate to predict anti-angiogenic (AA) response. However, the clinical adoption of transcriptomic assays faces several challenges including standardization, time delay, and high cost. Further, ccRCC tumors are highly heterogenous, and sampling multiple areas for sequencing is impractical. Here we present a novel deep learning (DL) approach to predict the Angioscore from ubiquitous histopathology slides. To overcome the lack of interpretability, one of the biggest limitations of typical DL models, our model produces a visual vascular network which is the basis of the model's prediction. To test its reliability, we applied this model to multiple cohorts including a clinical trial dataset. Our model accurately predicts the RNA-based Angioscore on multiple independent cohorts (spearman correlations of 0.77 and 0.73). Further, the predictions help unravel meaningful biology such as association of angiogenesis with grade, stage, and driver mutation status. Finally, we find our model can predict response to AA therapy, in both a real-world cohort and the IMmotion150 clinical trial. The predictive power of our model vastly exceeds that of CD31, a marker of vasculature, and nearly rivals the performance (c-index 0.66 vs 0.67) of the ground truth RNA-based Angioscore at a fraction of the cost. By providing a robust yet interpretable prediction of the Angioscore from histopathology slides alone, our approach offers insights into angiogenesis biology and AA treatment response.
Authors: Renzhi Wu, Pramod Chunduri, Dristi J Shah, Ashmitha Julius Aravind, Ali Payani, Xu Chu, Joy Arulraj, Kexin Rong
Abstract: In this paper, we will present SketchQL, a video database management system (VDBMS) for retrieving video moments with a sketch-based query interface. This novel interface allows users to specify object trajectory events with simple mouse drag-and-drop operations. Users can use trajectories of single objects as building blocks to compose complex events. Using a pre-trained model that encodes trajectory similarity, SketchQL achieves zero-shot video moments retrieval by performing similarity searches over the video to identify clips that are the most similar to the visual query. In this demonstration, we introduce the graphic user interface of SketchQL and detail its functionalities and interaction mechanisms. We also demonstrate the end-to-end usage of SketchQL from query composition to video moments retrieval using real-world scenarios.
Authors: Silvia Garc\'ia M\'endez, F\'atima Leal, Benedita Malheiro, Juan Carlos Burguillo Rial
Abstract: Wiki articles are created and maintained by a crowd of editors, producing a continuous stream of reviews. Reviews can take the form of additions, reverts, or both. This crowdsourcing model is exposed to manipulation since neither reviews nor editors are automatically screened and purged. To protect articles against vandalism or damage, the stream of reviews can be mined to classify reviews and profile editors in real-time. The goal of this work is to anticipate and explain which reviews to revert. This way, editors are informed why their edits will be reverted. The proposed method employs stream-based processing, updating the profiling and classification models on each incoming event. The profiling uses side and content-based features employing Natural Language Processing, and editor profiles are incrementally updated based on their reviews. Since the proposed method relies on self-explainable classification algorithms, it is possible to understand why a review has been classified as a revert or a non-revert. In addition, this work contributes an algorithm for generating synthetic data for class balancing, making the final classification fairer. The proposed online method was tested with a real data set from Wikivoyage, which was balanced through the aforementioned synthetic data generation. The results attained near-90 % values for all evaluation metrics (accuracy, precision, recall, and F-measure).
Authors: Somnath Kumar, Yash Gadhia, Tanuja Ganu, Akshay Nambi
Abstract: Recent advancements in Multi-modal Large Language Models (MLLMs) have significantly improved their performance in tasks combining vision and language. However, challenges persist in detailed multi-modal understanding, comprehension of complex tasks, and reasoning over multi-modal information. This paper introduces MMCTAgent, a novel multi-modal critical thinking agent framework designed to address the inherent limitations of current MLLMs in complex visual reasoning tasks. Inspired by human cognitive processes and critical thinking, MMCTAgent iteratively analyzes multi-modal information, decomposes queries, plans strategies, and dynamically evolves its reasoning. Additionally, MMCTAgent incorporates critical thinking elements such as verification of final answers and self-reflection through a novel approach that defines a vision-based critic and identifies task-specific evaluation criteria, thereby enhancing its decision-making abilities. Through rigorous evaluations across various image and video understanding benchmarks, we demonstrate that MMCTAgent (with and without the critic) outperforms both foundational MLLMs and other tool-augmented pipelines.
Authors: Somnath Kumar, Vaibhav Balloli, Mercy Ranjit, Kabir Ahuja, Tanuja Ganu, Sunayana Sitaram, Kalika Bali, Akshay Nambi
Abstract: Large language models (LLMs) are at the forefront of transforming numerous domains globally. However, their inclusivity and effectiveness remain limited for non-Latin scripts and low-resource languages. This paper tackles the imperative challenge of enhancing the multilingual performance of LLMs without extensive training or fine-tuning. Through systematic investigation and evaluation of diverse languages using popular question-answering (QA) datasets, we present novel techniques that unlock the true potential of LLMs in a polyglot landscape. Our approach encompasses three key strategies that yield significant improvements in multilingual proficiency. First, by meticulously optimizing prompts tailored for polyglot LLMs, we unlock their latent capabilities, resulting in substantial performance boosts across languages. Second, we introduce a new hybrid approach that synergizes LLM Retrieval Augmented Generation (RAG) with multilingual embeddings and achieves improved multilingual task performance. Finally, we introduce a novel learning approach that dynamically selects the optimal prompt strategy, LLM model, and embedding model per query at run-time. This dynamic adaptation maximizes the efficacy of LLMs across languages, outperforming best static and random strategies. Additionally, our approach adapts configurations in both offline and online settings, and can seamlessly adapt to new languages and datasets, leading to substantial advancements in multilingual understanding and generation across diverse languages.
Authors: Eshaan Agarwal, Vivek Dani, Tanuja Ganu, Akshay Nambi
Abstract: Large language models (LLMs) have revolutionized AI across diverse domains, showcasing remarkable capabilities. Central to their success is the concept of prompting, which guides model output generation. However, manual prompt engineering is labor-intensive and domain-specific, necessitating automated solutions. This paper introduces PromptWizard, a novel framework leveraging LLMs to iteratively synthesize and refine prompts tailored to specific tasks. Unlike existing approaches, PromptWizard optimizes both prompt instructions and in-context examples, maximizing model performance. The framework iteratively refines prompts by mutating instructions and incorporating negative examples to deepen understanding and ensure diversity. It further enhances both instructions and examples with the aid of a critic, synthesizing new instructions and examples enriched with detailed reasoning steps for optimal performance. PromptWizard offers several key features and capabilities, including computational efficiency compared to state-of-the-art approaches, adaptability to scenarios with varying amounts of training data, and effectiveness with smaller LLMs. Rigorous evaluation across 35 tasks on 8 datasets demonstrates PromptWizard's superiority over existing prompt strategies, showcasing its efficacy and scalability in prompt optimization.
Authors: Xiang Li, Zebang Shen, Liang Zhang, Niao He
Abstract: Continuous-time approximation of Stochastic Gradient Descent (SGD) is a crucial tool to study its escaping behaviors from stationary points. However, existing stochastic differential equation (SDE) models fail to fully capture these behaviors, even for simple quadratic objectives. Built on a novel stochastic backward error analysis framework, we derive the Hessian-Aware Stochastic Modified Equation (HA-SME), an SDE that incorporates Hessian information of the objective function into both its drift and diffusion terms. Our analysis shows that HA-SME matches the order-best approximation error guarantee among existing SDE models in the literature, while achieving a significantly reduced dependence on the smoothness parameter of the objective. Further, for quadratic objectives, under mild conditions, HA-SME is proved to be the first SDE model that recovers exactly the SGD dynamics in the distributional sense. Consequently, when the local landscape near a stationary point can be approximated by quadratics, HA-SME is expected to accurately predict the local escaping behaviors of SGD.
Authors: Tijana Zrnic
Abstract: We introduce PPBoot: a bootstrap-based method for prediction-powered inference. PPBoot is applicable to arbitrary estimation problems and is very simple to implement, essentially only requiring one application of the bootstrap. Through a series of examples, we demonstrate that PPBoot often performs nearly identically to (and sometimes better than) the earlier PPI(++) method based on asymptotic normality$\unicode{x2013}$when the latter is applicable$\unicode{x2013}$without requiring any asymptotic characterizations. Given its versatility, PPBoot could simplify and expand the scope of application of prediction-powered inference to problems where central limit theorems are hard to prove.
Authors: Dominic LaBella, Katherine Schumacher, Michael Mix, Kevin Leu, Shan McBurney-Lin, Pierre Nedelec, Javier Villanueva-Meyer, Jonathan Shapey, Tom Vercauteren, Kazumi Chia, Omar Al-Salihi, Justin Leu, Lia Halasz, Yury Velichko, Chunhao Wang, John Kirkpatrick, Scott Floyd, Zachary J. Reitman, Trey Mullikin, Ulas Bagci, Sean Sachdev, Jona A. Hattangadi-Gluth, Tyler Seibert, Nikdokht Farid, Connor Puett, Matthew W. Pease, Kevin Shiue, Syed Muhammad Anwar, Shahriar Faghani, Muhammad Ammar Haider, Pranav Warman, Jake Albrecht, Andr\'as Jakab, Mana Moassefi, Verena Chung, Alejandro Aristizabal, Alexandros Karargyris, Hasan Kassem, Sarthak Pati, Micah Sheller, Christina Huang, Aaron Coley, Siddharth Ghanta, Alex Schneider, Conrad Sharp, Rachit Saluja, Florian Kofler, Philipp Lohmann, Phillipp Vollmuth, Louis Gagnon, Maruf Adewole, Hongwei Bran Li, Anahita Fathi Kazerooni, Nourel Hoda Tahon, Udunna Anazodo, Ahmed W. Moawad, Bjoern Menze, Marius George Linguraru, Mariam Aboian, Benedikt Wiestler, Ujjwal Baid, Gian-Marco Conte, Andreas M. T. Rauschecker, Ayman Nada, Aly H. Abayazeed, Raymond Huang, Maria Correia de Verdier, Jeffrey D. Rudie, Spyridon Bakas, Evan Calabrese
Abstract: The 2024 Brain Tumor Segmentation Meningioma Radiotherapy (BraTS-MEN-RT) challenge aims to advance automated segmentation algorithms using the largest known multi-institutional dataset of radiotherapy planning brain MRIs with expert-annotated target labels for patients with intact or post-operative meningioma that underwent either conventional external beam radiotherapy or stereotactic radiosurgery. Each case includes a defaced 3D post-contrast T1-weighted radiotherapy planning MRI in its native acquisition space, accompanied by a single-label "target volume" representing the gross tumor volume (GTV) and any at-risk post-operative site. Target volume annotations adhere to established radiotherapy planning protocols, ensuring consistency across cases and institutions. For pre-operative meningiomas, the target volume encompasses the entire GTV and associated nodular dural tail, while for post-operative cases, it includes at-risk resection cavity margins as determined by the treating institution. Case annotations were reviewed and approved by expert neuroradiologists and radiation oncologists. Participating teams will develop, containerize, and evaluate automated segmentation models using this comprehensive dataset. Model performance will be assessed using the lesion-wise Dice Similarity Coefficient and the 95% Hausdorff distance. The top-performing teams will be recognized at the Medical Image Computing and Computer Assisted Intervention Conference in October 2024. BraTS-MEN-RT is expected to significantly advance automated radiotherapy planning by enabling precise tumor segmentation and facilitating tailored treatment, ultimately improving patient outcomes.
Authors: Yixiao Zhang, Yukara Ikemiya, Woosung Choi, Naoki Murata, Marco A. Mart\'inez-Ram\'irez, Liwei Lin, Gus Xia, Wei-Hsiang Liao, Yuki Mitsufuji, Simon Dixon
Abstract: Recent advances in text-to-music editing, which employ text queries to modify music (e.g.\ by changing its style or adjusting instrumental components), present unique challenges and opportunities for AI-assisted music creation. Previous approaches in this domain have been constrained by the necessity to train specific editing models from scratch, which is both resource-intensive and inefficient; other research uses large language models to predict edited music, resulting in imprecise audio reconstruction. To Combine the strengths and address these limitations, we introduce Instruct-MusicGen, a novel approach that finetunes a pretrained MusicGen model to efficiently follow editing instructions such as adding, removing, or separating stems. Our approach involves a modification of the original MusicGen architecture by incorporating a text fusion module and an audio fusion module, which allow the model to process instruction texts and audio inputs concurrently and yield the desired edited music. Remarkably, Instruct-MusicGen only introduces 8% new parameters to the original MusicGen model and only trains for 5K steps, yet it achieves superior performance across all tasks compared to existing baselines, and demonstrates performance comparable to the models trained for specific tasks. This advancement not only enhances the efficiency of text-to-music editing but also broadens the applicability of music language models in dynamic music production environments.
Authors: Ethan Shen, Alan Fan, Sarah M Pratt, Jae Sung Park, Matthew Wallingford, Sham M. Kakade, Ari Holtzman, Ranjay Krishna, Ali Farhadi, Aditya Kusupati
Abstract: Many applications today provide users with multiple auto-complete drafts as they type, including GitHub's code completion, Gmail's smart compose, and Apple's messaging auto-suggestions. Under the hood, language models support this by running an autoregressive inference pass to provide a draft. Consequently, providing $k$ drafts to the user requires running an expensive language model $k$ times. To alleviate the computation cost of running $k$ inference passes, we propose Superposed Decoding, a new decoding algorithm that generates $k$ drafts at the computation cost of one autoregressive inference pass. We achieve this by feeding a superposition of the $k$ most recent token embeddings from the drafts as input to the next decoding step of the language model. At every inference step we combine the $k$ drafts with the top-$k$ tokens to get $k^2$ new drafts and cache the $k$ most likely options, using an n-gram interpolation with minimal compute overhead to filter out incoherent generations. Our experiments show that $k$ drafts from Superposed Decoding are at least as coherent and factual as Nucleus Sampling and Greedy Decoding respectively, while being at least $2.44\times$ faster for $k\ge3$. In a compute-normalized setting, user evaluations demonstrably favor text generated by Superposed Decoding over Nucleus Sampling. Code and more examples open-sourced at https://github.com/RAIVNLab/SuperposedDecoding.
Authors: Jialin Dong, Bahare Fatemi, Bryan Perozzi, Lin F. Yang, Anton Tsitsulin
Abstract: Retrieval Augmented Generation (RAG) has greatly improved the performance of Large Language Model (LLM) responses by grounding generation with context from existing documents. These systems work well when documents are clearly relevant to a question context. But what about when a document has partial information, or less obvious connections to the context? And how should we reason about connections between documents? In this work, we seek to answer these two core questions about RAG generation. We introduce G-RAG, a reranker based on graph neural networks (GNNs) between the retriever and reader in RAG. Our method combines both connections between documents and semantic information (via Abstract Meaning Representation graphs) to provide a context-informed ranker for RAG. G-RAG outperforms state-of-the-art approaches while having smaller computational footprint. Additionally, we assess the performance of PaLM 2 as a reranker and find it to significantly underperform G-RAG. This result emphasizes the importance of reranking for RAG even when using Large Language Models.
Authors: Yuhui Zhang, Alyssa Unell, Xiaohan Wang, Dhruba Ghosh, Yuchang Su, Ludwig Schmidt, Serena Yeung-Levy
Abstract: Image classification is one of the most fundamental capabilities of machine vision intelligence. In this work, we revisit the image classification task using visually-grounded language models (VLMs) such as GPT-4V and LLaVA. We find that existing proprietary and public VLMs, despite often using CLIP as a vision encoder and having many more parameters, significantly underperform CLIP on standard image classification benchmarks like ImageNet. To understand the reason, we explore several hypotheses concerning the inference algorithms, training objectives, and data processing in VLMs. Our analysis reveals that the primary cause is data-related: critical information for image classification is encoded in the VLM's latent space but can only be effectively decoded with enough training data. Specifically, there is a strong correlation between the frequency of class exposure during VLM training and instruction-tuning and the VLM's performance in those classes; when trained with sufficient data, VLMs can match the accuracy of state-of-the-art classification models. Based on these findings, we enhance a VLM by integrating classification-focused datasets into its training, and demonstrate that the enhanced classification performance of the VLM transfers to its general capabilities, resulting in an improvement of 11.8% on the newly collected ImageWikiQA dataset.
Authors: Khen Cohen, Noam Levi, Yaron Oz
Abstract: We derive closed-form expressions for the Bayes optimal decision boundaries in binary classification of high dimensional overlapping Gaussian mixture model (GMM) data, and show how they depend on the eigenstructure of the class covariances, for particularly interesting structured data. We empirically demonstrate, through experiments on synthetic GMMs inspired by real-world data, that deep neural networks trained for classification, learn predictors which approximate the derived optimal classifiers. We further extend our study to networks trained on authentic data, observing that decision thresholds correlate with the covariance eigenvectors rather than the eigenvalues, mirroring our GMM analysis. This provides theoretical insights regarding neural networks' ability to perform probabilistic inference and distill statistical patterns from intricate distributions.
Authors: Elisabeth Ailer, Christian L. M\"uller, Niki Kilbertus
Abstract: Many scientific datasets are compositional in nature. Important biological examples include species abundances in ecology, cell-type compositions derived from single-cell sequencing data, and amplicon abundance data in microbiome research. Here, we provide a causal view on compositional data in an instrumental variable setting where the composition acts as the cause. First, we crisply articulate potential pitfalls for practitioners regarding the interpretation of compositional causes from the viewpoint of interventions and warn against attributing causal meaning to common summary statistics such as diversity indices in microbiome data analysis. We then advocate for and develop multivariate methods using statistical data transformations and regression techniques that take the special structure of the compositional sample space into account while still yielding scientifically interpretable results. In a comparative analysis on synthetic and real microbiome data we show the advantages and limitations of our proposal. We posit that our analysis provides a useful framework and guidance for valid and informative cause-effect estimation in the context of compositional data.
Authors: Albin Soutif--Cormerais, Marc Masana, Joost van de Weijer, Bart{\l}omiej Twardowski
Abstract: In class-incremental learning, an agent with limited resources needs to learn a sequence of classification tasks, forming an ever growing classification problem, with the constraint of not being able to access data from previous tasks. The main difference with task-incremental learning, where a task-ID is available at inference time, is that the learner also needs to perform cross-task discrimination, i.e. distinguish between classes that have not been seen together. Approaches to tackle this problem are numerous and mostly make use of an external memory (buffer) of non-negligible size. In this paper, we ablate the learning of cross-task features and study its influence on the performance of basic replay strategies used for class-IL. We also define a new forgetting measure for class-incremental learning, and see that forgetting is not the principal cause of low performance. Our experimental results show that future algorithms for class-incremental learning should not only prevent forgetting, but also aim to improve the quality of the cross-task features, and the knowledge transfer between tasks. This is especially important when tasks contain limited amount of data.
Authors: Sergio Rozada, Santiago Paternain, Antonio G. Marques
Abstract: Value-function (VF) approximation is a central problem in Reinforcement Learning (RL). Classical non-parametric VF estimation suffers from the curse of dimensionality. As a result, parsimonious parametric models have been adopted to approximate VFs in high-dimensional spaces, with most efforts being focused on linear and neural-network-based approaches. Differently, this paper puts forth a a parsimonious non-parametric approach, where we use stochastic low-rank algorithms to estimate the VF matrix in an online and model-free fashion. Furthermore, as VFs tend to be multi-dimensional, we propose replacing the classical VF matrix representation with a tensor (multi-way array) representation and, then, use the PARAFAC decomposition to design an online model-free tensor low-rank algorithm. Different versions of the algorithms are proposed, their complexity is analyzed, and their performance is assessed numerically using standardized RL environments.
Authors: Katharina Rombach, Gabriel Michau, Wilfried B\"urzle, Stefan Koller, Olga Fink
Abstract: Monitoring the health of complex industrial assets is crucial for safe and efficient operations. Health indicators that provide quantitative real-time insights into the health status of industrial assets over time serve as valuable tools for e.g. fault detection or prognostics. This study proposes a novel, versatile and unsupervised approach to learn health indicators using contrastive learning, where the operational time serves as a proxy for degradation. To highlight its versatility, the approach is evaluated on two tasks and case studies with different characteristics: wear assessment of milling machines and fault detection of railway wheels. Our results show that the proposed methodology effectively learns a health indicator that follows the wear of milling machines (0.97 correlation on average) and is suitable for fault detection in railway wheels (88.7% balanced accuracy). The conducted experiments demonstrate the versatility of the approach for various systems and health conditions.
Authors: Zahra Zamanzadeh Darban, Geoffrey I. Webb, Shirui Pan, Charu C. Aggarwal, Mahsa Salehi
Abstract: Time series anomaly detection has applications in a wide range of research fields and applications, including manufacturing and healthcare. The presence of anomalies can indicate novel or unexpected events, such as production faults, system defects, or heart fluttering, and is therefore of particular interest. The large size and complex patterns of time series have led researchers to develop specialised deep learning models for detecting anomalous patterns. This survey focuses on providing structured and comprehensive state-of-the-art time series anomaly detection models through the use of deep learning. It providing a taxonomy based on the factors that divide anomaly detection models into different categories. Aside from describing the basic anomaly detection technique for each category, the advantages and limitations are also discussed. Furthermore, this study includes examples of deep anomaly detection in time series across various application domains in recent years. It finally summarises open issues in research and challenges faced while adopting deep anomaly detection models.
Authors: Yun-Wei Chu, Seyyedali Hosseinalipour, Elizabeth Tenorio, Laura Cruz, Kerrie Douglas, Andrew Lan, Christopher Brinton
Abstract: Conventional methods for student modeling, which involve predicting grades based on measured activities, struggle to provide accurate results for minority/underrepresented student groups due to data availability biases. In this paper, we propose a Multi-Layer Personalized Federated Learning (MLPFL) methodology that optimizes inference accuracy over different layers of student grouping criteria, such as by course and by demographic subgroups within each course. In our approach, personalized models for individual student subgroups are derived from a global model, which is trained in a distributed fashion via meta-gradient updates that account for subgroup heterogeneity while preserving modeling commonalities that exist across the full dataset. The evaluation of the proposed methodology considers case studies of two popular downstream student modeling tasks, knowledge tracing and outcome prediction, which leverage multiple modalities of student behavior (e.g., visits to lecture videos and participation on forums) in model training. Experiments on three real-world online course datasets show significant improvements achieved by our approach over existing student modeling benchmarks, as evidenced by an increased average prediction quality and decreased variance across different student subgroups. Visual analysis of the resulting students' knowledge state embeddings confirm that our personalization methodology extracts activity patterns clustered into different student subgroups, consistent with the performance enhancements we obtain over the baselines.
Authors: Ha Manh Bui, Anqi Liu
Abstract: Sampling-based methods, e.g., Deep Ensembles and Bayesian Neural Nets have become promising approaches to improve the quality of uncertainty estimation and robust generalization. However, they suffer from a large model size and high latency at test-time, which limits the scalability needed for low-resource devices and real-time applications. To resolve these computational issues, we propose Density-Softmax, a sampling-free deterministic framework via combining a density function built on a Lipschitz-constrained feature extractor with the softmax layer. Theoretically, we show that our model is the solution of minimax uncertainty risk and is distance-aware on feature space, thus reducing the over-confidence of the standard softmax under distribution shifts. Empirically, our method enjoys competitive results with state-of-the-art techniques in terms of uncertainty and robustness, while having a lower number of model parameters and a lower latency at test-time.
Authors: Marco Mussi, Alessandro Montenegro, Francesco Trov\'o, Marcello Restelli, Alberto Maria Metelli
Abstract: Stochastic Rising Bandits (SRBs) model sequential decision-making problems in which the expected reward of the available options increases every time they are selected. This setting captures a wide range of scenarios in which the available options are learning entities whose performance improves (in expectation) over time (e.g., online best model selection). While previous works addressed the regret minimization problem, this paper focuses on the fixed-budget Best Arm Identification (BAI) problem for SRBs. In this scenario, given a fixed budget of rounds, we are asked to provide a recommendation about the best option at the end of the identification process. We propose two algorithms to tackle the above-mentioned setting, namely R-UCBE, which resorts to a UCB-like approach, and R-SR, which employs a successive reject procedure. Then, we prove that, with a sufficiently large budget, they provide guarantees on the probability of properly identifying the optimal option at the end of the learning process and on the simple regret. Furthermore, we derive a lower bound on the error probability, matched by our R-SR (up to constants), and illustrate how the need for a sufficiently large budget is unavoidable in the SRB setting. Finally, we numerically validate the proposed algorithms in both synthetic and realistic environments.
Authors: Song Bian, Xiating Ouyang, Zhiwei Fan, Paraschos Koutris
Abstract: We study the certifiable robustness of ML classifiers on dirty datasets that could contain missing values. A test point is certifiably robust for an ML classifier if the classifier returns the same prediction for that test point, regardless of which cleaned version (among exponentially many) of the dirty dataset the classifier is trained on. In this paper, we show theoretically that for Naive Bayes Classifiers (NBC) over dirty datasets with missing values: (i) there exists an efficient polynomial time algorithm to decide whether multiple input test points are all certifiably robust over a dirty dataset; and (ii) the data poisoning attack, which aims to make all input test points certifiably non-robust by inserting missing cells to the clean dataset, is in polynomial time for single test points but NP-complete for multiple test points. Extensive experiments demonstrate that our algorithms are efficient and outperform existing baselines.
Authors: Yumeng Song, Yu Gu, Tianyi Li, Jianzhong Qi, Zhenghao Liu, Christian S. Jensen, Ge Yu
Abstract: Hypergraphs can model higher-order relationships among data objects that are found in applications such as social networks and bioinformatics. However, recent studies on hypergraph learning that extend graph convolutional networks to hypergraphs cannot learn effectively from features of unlabeled data. To such learning, we propose a contrastive hypergraph neural network, CHGNN, that exploits self-supervised contrastive learning techniques to learn from labeled and unlabeled data. First, CHGNN includes an adaptive hypergraph view generator that adopts an auto-augmentation strategy and learns a perturbed probability distribution of minimal sufficient views. Second, CHGNN encompasses an improved hypergraph encoder that considers hyperedge homogeneity to fuse information effectively. Third, CHGNN is equipped with a joint loss function that combines a similarity loss for the view generator, a node classification loss, and a hyperedge homogeneity loss to inject supervision signals. It also includes basic and cross-validation contrastive losses, associated with an enhanced contrastive loss training process. Experimental results on nine real datasets offer insight into the effectiveness of CHGNN, showing that it outperforms 13 competitors in terms of classification accuracy consistently.
Authors: Dong Wang, Olga Saukh, Xiaoxi He, Lothar Thiele
Abstract: While the deployment of deep learning models on edge devices is increasing, these models often lack robustness when faced with dynamic changes in sensed data. This can be attributed to sensor drift, or variations in the data compared to what was used during offline training due to factors such as specific sensor placement or naturally changing sensing conditions. Hence, achieving the desired robustness necessitates the utilization of either an invariant architecture or specialized training approaches, like data augmentation techniques. Alternatively, input transformations can be treated as a domain shift problem, and solved by post-deployment model adaptation. In this paper, we train a parameterized subspace of configurable networks, where an optimal network for a particular parameter setting is part of this subspace. The obtained subspace is low-dimensional and has a surprisingly simple structure even for complex, non-invertible transformations of the input, leading to an exceptionally high efficiency of subspace-configurable networks (SCNs) when limited storage and computing resources are at stake.
Authors: Yibo Miao, Hongcheng Gao, Hao Zhang, Zhijie Deng
Abstract: The detection of machine-generated text, especially from large language models (LLMs), is crucial in preventing serious social problems resulting from their misuse. Some methods train dedicated detectors on specific datasets but fall short in generalizing to unseen test data, while other zero-shot ones often yield suboptimal performance. Although the recent DetectGPT has shown promising detection performance, it suffers from significant inefficiency issues, as detecting a single candidate requires querying the source LLM with hundreds of its perturbations. This paper aims to bridge this gap. Concretely, we propose to incorporate a Bayesian surrogate model, which allows us to select typical samples based on Bayesian uncertainty and interpolate scores from typical samples to other samples, to improve query efficiency. Empirical results demonstrate that our method significantly outperforms existing approaches under a low query budget. Notably, when detecting the text generated by LLaMA family models, our method with just 2 or 3 queries can outperform DetectGPT with 200 queries.
Authors: Licong Lin, Tijana Zrnic
Abstract: When predictions are performative, the choice of which predictor to deploy influences the distribution of future observations. The overarching goal in learning under performativity is to find a predictor that has low \emph{performative risk}, that is, good performance on its induced distribution. One family of solutions for optimizing the performative risk, including bandits and other derivative-free methods, is agnostic to any structure in the performative feedback, leading to exceedingly slow convergence rates. A complementary family of solutions makes use of explicit \emph{models} for the feedback, such as best-response models in strategic classification, enabling faster rates. However, these rates critically rely on the feedback model being correct. In this work we study a general protocol for making use of possibly misspecified models in performative prediction, called \emph{plug-in performative optimization}. We show this solution can be far superior to model-agnostic strategies, as long as the misspecification is not too extreme. Our results support the hypothesis that models, even if misspecified, can indeed help with learning in performative settings.
Authors: Vinitra Swamy, Jibril Frej, Tanja K\"aser
Abstract: Explainable Artificial Intelligence (XAI) plays a crucial role in enabling human understanding and trust in deep learning systems. As models get larger, more ubiquitous, and pervasive in aspects of daily life, explainability is necessary to minimize adverse effects of model mistakes. Unfortunately, current approaches in human-centric XAI (e.g. predictive tasks in healthcare, education, or personalized ads) tend to rely on a single post-hoc explainer, whereas recent work has identified systematic disagreement between post-hoc explainers when applied to the same instances of underlying black-box models. In this paper, we therefore present a call for action to address the limitations of current state-of-the-art explainers. We propose a shift from post-hoc explainability to designing interpretable neural network architectures. We identify five needs of human-centric XAI (real-time, accurate, actionable, human-interpretable, and consistent) and propose two schemes for interpretable-by-design neural network workflows (adaptive routing with InterpretCC and temporal diagnostics with I2MD). We postulate that the future of human-centric XAI is neither in explaining black-boxes nor in reverting to traditional, interpretable models, but in neural networks that are intrinsically interpretable.
Authors: Edward Pearce-Crump, William J. Knottenbelt
Abstract: Permutation equivariant neural networks are typically used to learn from data that lives on a graph. However, for any graph $G$ that has $n$ vertices, using the symmetric group $S_n$ as its group of symmetries does not take into account the relations that exist between the vertices. Given that the actual group of symmetries is the automorphism group Aut$(G)$, we show how to construct neural networks that are equivariant to Aut$(G)$ by obtaining a full characterisation of the learnable, linear, Aut$(G)$-equivariant functions between layers that are some tensor power of $\mathbb{R}^{n}$. In particular, we find a spanning set of matrices for these layer functions in the standard basis of $\mathbb{R}^{n}$. This result has important consequences for learning from data whose group of symmetries is a finite group because a theorem by Frucht (1938) showed that any finite group is isomorphic to the automorphism group of a graph.
Authors: Saeed Masoudian, Julian Zimmert, Yevgeny Seldin
Abstract: We propose a new best-of-both-worlds algorithm for bandits with variably delayed feedback. In contrast to prior work, which required prior knowledge of the maximal delay $d_{\mathrm{max}}$ and had a linear dependence of the regret on it, our algorithm can tolerate arbitrary excessive delays up to order $T$ (where $T$ is the time horizon). The algorithm is based on three technical innovations, which may all be of independent interest: (1) We introduce the first implicit exploration scheme that works in best-of-both-worlds setting. (2) We introduce the first control of distribution drift that does not rely on boundedness of delays. The control is based on the implicit exploration scheme and adaptive skipping of observations with excessive delays. (3) We introduce a procedure relating standard regret with drifted regret that does not rely on boundedness of delays. At the conceptual level, we demonstrate that complexity of best-of-both-worlds bandits with delayed feedback is characterized by the amount of information missing at the time of decision making (measured by the number of outstanding observations) rather than the time that the information is missing (measured by the delays).
Authors: Nikita Balabin, Daria Voronkova, Ilya Trofimov, Evgeny Burnaev, Serguei Barannikov
Abstract: We propose TopDis (Topological Disentanglement), a method for learning disentangled representations via adding a multi-scale topological loss term. Disentanglement is a crucial property of data representations substantial for the explainability and robustness of deep learning models and a step towards high-level cognition. The state-of-the-art methods are based on VAE and encourage the joint distribution of latent variables to be factorized. We take a different perspective on disentanglement by analyzing topological properties of data manifolds. In particular, we optimize the topological similarity for data manifolds traversals. To the best of our knowledge, our paper is the first one to propose a differentiable topological loss for disentanglement learning. Our experiments have shown that the proposed TopDis loss improves disentanglement scores such as MIG, FactorVAE score, SAP score, and DCI disentanglement score with respect to state-of-the-art results while preserving the reconstruction quality. Our method works in an unsupervised manner, permitting us to apply it to problems without labeled factors of variation. The TopDis loss works even when factors of variation are correlated. Additionally, we show how to use the proposed topological loss to find disentangled directions in a trained GAN.
Authors: Yuancheng Xu, Chenghao Deng, Yanchao Sun, Ruijie Zheng, Xiyao Wang, Jieyu Zhao, Furong Huang
Abstract: Decisions made by machine learning models can have lasting impacts, making long-term fairness a critical consideration. It has been observed that ignoring the long-term effect and directly applying fairness criterion in static settings can actually worsen bias over time. To address biases in sequential decision-making, we introduce a long-term fairness concept named Equal Long-term Benefit Rate (ELBERT). This concept is seamlessly integrated into a Markov Decision Process (MDP) to consider the future effects of actions on long-term fairness, thus providing a unified framework for fair sequential decision-making problems. ELBERT effectively addresses the temporal discrimination issues found in previous long-term fairness notions. Additionally, we demonstrate that the policy gradient of Long-term Benefit Rate can be analytically simplified to standard policy gradients. This simplification makes conventional policy optimization methods viable for reducing bias, leading to our bias mitigation approach ELBERT-PO. Extensive experiments across various diverse sequential decision-making environments consistently reveal that ELBERT-PO significantly diminishes bias while maintaining high utility. Code is available at https://github.com/umd-huang-lab/ELBERT.
Authors: Khang Ly, Yury Kashnitsky, Savvas Chamezopoulos, Valeria Krzhizhanovskaya
Abstract: Classifying research output into context-specific label taxonomies is a challenging and relevant downstream task, given the volume of existing and newly published articles. We propose a method to enhance the performance of article classification by enriching simple Graph Neural Network (GNN) pipelines with multi-graph representations that simultaneously encode multiple signals of article relatedness, e.g. references, co-authorship, shared publication source, shared subject headings, as distinct edge types. Fully supervised transductive node classification experiments are conducted on the Open Graph Benchmark OGBN-arXiv dataset and the PubMed diabetes dataset, augmented with additional metadata from Microsoft Academic Graph and PubMed Central, respectively. The results demonstrate that multi-graphs consistently improve the performance of a variety of GNN models compared to the default graphs. When deployed with SOTA textual node embedding methods, the transformed multi-graphs enable simple and shallow 2-layer GNN pipelines to achieve results on par with more complex architectures.
Authors: Flavio Figueiredo, Jos\'e Geraldo Fernandes, Jackson Silva, Renato M. Assun\c{c}\~ao
Abstract: Copulas are powerful statistical tools for capturing dependencies across data dimensions. Applying Copulas involves estimating independent marginals, a straightforward task, followed by the much more challenging task of determining a single copulating function, $C$, that links these marginals. For bivariate data, a copula takes the form of a two-increasing function $C: (u,v)\in \mathbb{I}^2 \rightarrow \mathbb{I}$, where $\mathbb{I} = [0, 1]$. This paper proposes 2-Cats, a Neural Network (NN) model that learns two-dimensional Copulas without relying on specific Copula families (e.g., Archimedean). Furthermore, via both theoretical properties of the model and a Lagrangian training approach, we show that 2-Cats meets the desiderata of Copula properties. Moreover, inspired by the literature on Physics-Informed Neural Networks and Sobolev Training, we further extend our training strategy to learn not only the output of a Copula but also its derivatives. Our proposed method exhibits superior performance compared to the state-of-the-art across various datasets while respecting (provably for most and approximately for a single other) properties of C.
Authors: Benjamin Hoover, Hendrik Strobelt, Dmitry Krotov, Judy Hoffman, Zsolt Kira, Duen Horng Chau
Abstract: The generative process of Diffusion Models (DMs) has recently set state-of-the-art on many AI generation benchmarks. Though the generative process is traditionally understood as an "iterative denoiser", there is no universally accepted language to describe it. We introduce a novel perspective to describe DMs using the mathematical language of memory retrieval from the field of energy-based Associative Memories (AMs), making efforts to keep our presentation approachable to newcomers to both of these fields. Unifying these two fields provides insight that DMs can be seen as a particular kind of AM where Lyapunov stability guarantees are bypassed by intelligently engineering the dynamics (i.e., the noise and step size schedules) of the denoising process. Finally, we present a growing body of evidence that records DMs exhibiting empirical behavior we would expect from AMs, and conclude by discussing research opportunities that are revealed by understanding DMs as a form of energy-based memory.
Authors: Emmeran Johnson, Ciara Pike-Burke, Patrick Rebeschini
Abstract: We theoretically explore the relationship between sample-efficiency and adaptivity in reinforcement learning. An algorithm is sample-efficient if it uses a number of queries $n$ to the environment that is polynomial in the dimension $d$ of the problem. Adaptivity refers to the frequency at which queries are sent and feedback is processed to update the querying strategy. To investigate this interplay, we employ a learning framework that allows sending queries in $K$ batches, with feedback being processed and queries updated after each batch. This model encompasses the whole adaptivity spectrum, ranging from non-adaptive 'offline' ($K=1$) to fully adaptive ($K=n$) scenarios, and regimes in between. For the problems of policy evaluation and best-policy identification under $d$-dimensional linear function approximation, we establish $\Omega(\log \log d)$ lower bounds on the number of batches $K$ required for sample-efficient algorithms with $n = O(poly(d))$ queries. Our results show that just having adaptivity ($K>1$) does not necessarily guarantee sample-efficiency. Notably, the adaptivity-boundary for sample-efficiency is not between offline reinforcement learning ($K=1$), where sample-efficiency was known to not be possible, and adaptive settings. Instead, the boundary lies between different regimes of adaptivity and depends on the problem dimension.
Authors: Enneng Yang, Zhenyi Wang, Li Shen, Shiwei Liu, Guibing Guo, Xingwei Wang, Dacheng Tao
Abstract: Multi-task learning (MTL) aims to empower a model to tackle multiple tasks simultaneously. A recent development known as task arithmetic has revealed that several models, each fine-tuned for distinct tasks, can be directly merged into a single model to execute MTL without necessitating a retraining process using the initial training data. Nevertheless, this direct addition of models often leads to a significant deterioration in the overall performance of the merged model. This decline occurs due to potential conflicts and intricate correlations among the multiple tasks. Consequently, the challenge emerges of how to merge pre-trained models more effectively without using their original training data. This paper introduces an innovative technique called Adaptive Model Merging (AdaMerging). This approach aims to autonomously learn the coefficients for model merging, either in a task-wise or layer-wise manner, without relying on the original training data. Specifically, our AdaMerging method operates as an automatic, unsupervised task arithmetic scheme. It leverages entropy minimization on unlabeled test samples from the multi-task setup as a surrogate objective function to iteratively refine the merging coefficients of the multiple models. Our experimental findings across eight tasks demonstrate the efficacy of the AdaMerging scheme we put forth. Compared to the current state-of-the-art task arithmetic merging scheme, AdaMerging showcases a remarkable 11\% improvement in performance. Notably, AdaMerging also exhibits superior generalization capabilities when applied to unseen downstream tasks. Furthermore, it displays a significantly enhanced robustness to data distribution shifts that may occur during the testing phase.
Authors: Azul Garza, Cristian Challu, Max Mergenthaler-Canseco
Abstract: In this paper, we introduce TimeGPT, the first foundation model for time series, capable of generating accurate predictions for diverse datasets not seen during training. We evaluate our pre-trained model against established statistical, machine learning, and deep learning methods, demonstrating that TimeGPT zero-shot inference excels in performance, efficiency, and simplicity. Our study provides compelling evidence that insights from other domains of artificial intelligence can be effectively applied to time series analysis. We conclude that large-scale time series models offer an exciting opportunity to democratize access to precise predictions and reduce uncertainty by leveraging the capabilities of contemporary advancements in deep learning.
Authors: Ke Wang, Guillermo Ortiz-Jimenez, Rodolphe Jenatton, Mark Collier, Efi Kokiopoulou, Pascal Frossard
Abstract: Label noise is a pervasive problem in deep learning that often compromises the generalization performance of trained models. Recently, leveraging privileged information (PI) -- information available only during training but not at test time -- has emerged as an effective approach to mitigate this issue. Yet, existing PI-based methods have failed to consistently outperform their no-PI counterparts in terms of preventing overfitting to label noise. To address this deficiency, we introduce Pi-DUAL, an architecture designed to harness PI to distinguish clean from wrong labels. Pi-DUAL decomposes the output logits into a prediction term, based on conventional input features, and a noise-fitting term influenced solely by PI. A gating mechanism steered by PI adaptively shifts focus between these terms, allowing the model to implicitly separate the learning paths of clean and wrong labels. Empirically, Pi-DUAL achieves significant performance improvements on key PI benchmarks (e.g., +6.8% on ImageNet-PI), establishing a new state-of-the-art test set accuracy. Additionally, Pi-DUAL is a potent method for identifying noisy samples post-training, outperforming other strong methods at this task. Overall, Pi-DUAL is a simple, scalable and practical approach for mitigating the effects of label noise in a variety of real-world scenarios with PI.
Authors: Luigi Sbail\`o, Luca Ghiringhelli
Abstract: We investigate the emergence of binary encoding within the latent space of deep-neural-network classifiers. Such binary encoding is induced by the introduction of a linear penultimate layer, which employs during training a loss function specifically designed to compress the latent representations. As a result of a trade-off between compression and information retention, the network learns to assume only one of two possible values for each dimension in the latent space. The binary encoding is provoked by the collapse of all representations of the same class to the same point, which corresponds to the vertex of a hypercube. By analyzing several datasets of increasing complexity, we provide empirical evidence that the emergence of binary encoding dramatically enhances robustness while also significantly improving the reliability and generalization of the network.
Authors: Dai Hai Nguyen, Tetsuya Sakurai, Hiroshi Mamitsuka
Abstract: Variational inference (VI) can be cast as an optimization problem in which the variational parameters are tuned to closely align a variational distribution with the true posterior. The optimization task can be approached through vanilla gradient descent in black-box VI or natural-gradient descent in natural-gradient VI. In this work, we reframe VI as the optimization of an objective that concerns probability distributions defined over a \textit{variational parameter space}. Subsequently, we propose Wasserstein gradient descent for tackling this optimization problem. Notably, the optimization techniques, namely black-box VI and natural-gradient VI, can be reinterpreted as specific instances of the proposed Wasserstein gradient descent. To enhance the efficiency of optimization, we develop practical methods for numerically solving the discrete gradient flows. We validate the effectiveness of the proposed methods through empirical experiments on a synthetic dataset, supplemented by theoretical analyses.
Authors: Jaiden Fairoze, Sanjam Garg, Somesh Jha, Saeed Mahloujifar, Mohammad Mahmoody, Mingyuan Wang
Abstract: We present a highly detectable, trustless watermarking scheme for LLMs: the detection algorithm contains no secret information, and it is executable by anyone. We embed a publicly-verifiable cryptographic signature into LLM output using rejection sampling. We prove that our scheme is cryptographically correct, sound, and distortion-free. We make novel uses of error-correction techniques to overcome periods of low entropy, a barrier for all prior watermarking schemes. We implement our scheme and make empirical measurements over open models in the 2.7B to 70B parameter range. Our experiments suggest that our formal claims are met in practice.
Authors: Ziqi Zhang, Xiao Xiong, Zifeng Zhuang, Jinxin Liu, Donglin Wang
Abstract: Studying how to fine-tune offline reinforcement learning (RL) pre-trained policy is profoundly significant for enhancing the sample efficiency of RL algorithms. However, directly fine-tuning pre-trained policies often results in sub-optimal performance. This is primarily due to the distribution shift between offline pre-training and online fine-tuning stages. Specifically, the distribution shift limits the acquisition of effective online samples, ultimately impacting the online fine-tuning performance. In order to narrow down the distribution shift between offline and online stages, we proposed Q conditioned state entropy (QCSE) as intrinsic reward. Specifically, QCSE maximizes the state entropy of all samples individually, considering their respective Q values. This approach encourages exploration of low-frequency samples while penalizing high-frequency ones, and implicitly achieves State Marginal Matching (SMM), thereby ensuring optimal performance, solving the asymptotic sub-optimality of constraint-based approaches. Additionally, QCSE can seamlessly integrate into various RL algorithms, enhancing online fine-tuning performance. To validate our claim, we conduct extensive experiments, and observe significant improvements with QCSE (about 13% for CQL and 8% for Cal-QL). Furthermore, we extended experimental tests to other algorithms, affirming the generality of QCSE.
Authors: Felix Chalumeau, Shikha Surana, Clement Bonnet, Nathan Grinsztajn, Arnu Pretorius, Alexandre Laterre, Thomas D. Barrett
Abstract: Combinatorial Optimization underpins many real-world applications and yet, designing performant algorithms to solve these complex, typically NP-hard, problems remains a significant research challenge. Reinforcement Learning (RL) provides a versatile framework for designing heuristics across a broad spectrum of problem domains. However, despite notable progress, RL has not yet supplanted industrial solvers as the go-to solution. Current approaches emphasize pre-training heuristics that construct solutions but often rely on search procedures with limited variance, such as stochastically sampling numerous solutions from a single policy or employing computationally expensive fine-tuning of the policy on individual problem instances. Building on the intuition that performant search at inference time should be anticipated during pre-training, we propose COMPASS, a novel RL approach that parameterizes a distribution of diverse and specialized policies conditioned on a continuous latent space. We evaluate COMPASS across three canonical problems - Travelling Salesman, Capacitated Vehicle Routing, and Job-Shop Scheduling - and demonstrate that our search strategy (i) outperforms state-of-the-art approaches on 11 standard benchmarking tasks and (ii) generalizes better, surpassing all other approaches on a set of 18 procedurally transformed instance distributions.
Authors: Derya Soydaner, Johan Wagemans
Abstract: The allure of aesthetic appeal in images captivates our senses, yet the underlying intricacies of aesthetic preferences remain elusive. In this study, we pioneer a novel perspective by utilizing several different machine learning (ML) models that focus on aesthetic attributes known to influence preferences. Our models process these attributes as inputs to predict the aesthetic scores of images. Moreover, to delve deeper and obtain interpretable explanations regarding the factors driving aesthetic preferences, we utilize the popular Explainable AI (XAI) technique known as SHapley Additive exPlanations (SHAP). Our methodology compares the performance of various ML models, including Random Forest, XGBoost, Support Vector Regression, and Multilayer Perceptron, in accurately predicting aesthetic scores, and consistently observing results in conjunction with SHAP. We conduct experiments on three image aesthetic benchmarks, namely Aesthetics with Attributes Database (AADB), Explainable Visual Aesthetics (EVA), and Personalized image Aesthetics database with Rich Attributes (PARA), providing insights into the roles of attributes and their interactions. Finally, our study presents ML models for aesthetics research, alongside the introduction of XAI. Our aim is to shed light on the complex nature of aesthetic preferences in images through ML and to provide a deeper understanding of the attributes that influence aesthetic judgements.
Authors: Yuli Slavutsky, Yuval Benjamini
Abstract: Zero-shot learning methods typically assume that the new, unseen classes that are encountered at deployment, come from the same distribution as training classes. However, real-world scenarios often involve class distribution shifts (e.g., in age or gender for person identification), posing challenges for zero-shot classifiers that rely on learned representations from training classes. In this work, we propose a model that assumes that the attribute responsible for the shift is unknown in advance, and show that standard training may lead to non-robust representations. To mitigate this, we propose an algorithm for learning robust representations by (a) constructing synthetic data environments via hierarchical sampling and (b) applying environment balancing penalization, inspired by out-of-distribution problems. We show that our approach improves generalization on diverse class distributions in both simulations and real-world datasets.
Authors: Xiaobei Zou, Luolin Xiong, Yang Tang, J\"urgen Kurths
Abstract: Spatio-temporal forecasting in various domains, like traffic prediction and weather forecasting, is a challenging endeavor, primarily due to the difficulties in modeling propagation dynamics and capturing high-dimensional interactions among nodes. Despite the significant strides made by graph-based networks in spatio-temporal forecasting, there remain two pivotal factors closely related to forecasting performance that need further consideration: time delays in propagation dynamics and multi-scale high-dimensional interactions. In this work, we present a Series-Aligned Multi-Scale Graph Learning (SAMSGL) framework, aiming to enhance forecasting performance. In order to handle time delays in spatial interactions, we propose a series-aligned graph convolution layer to facilitate the aggregation of non-delayed graph signals, thereby mitigating the influence of time delays for the improvement in accuracy. To understand global and local spatio-temporal interactions, we develop a spatio-temporal architecture via multi-scale graph learning, which encompasses two essential components: multi-scale graph structure learning and graph-fully connected (Graph-FC) blocks. The multi-scale graph structure learning includes a global graph structure to learn both delayed and non-delayed node embeddings, as well as a local one to learn node variations influenced by neighboring factors. The Graph-FC blocks synergistically fuse spatial and temporal information to boost prediction accuracy. To evaluate the performance of SAMSGL, we conduct experiments on meteorological and traffic forecasting datasets, which demonstrate its effectiveness and superiority.
Authors: Matthew L. Olson, Shusen Liu, Jayaraman J. Thiagarajan, Bogdan Kustowski, Weng-Keen Wong, Rushil Anirudh
Abstract: Recent advances in machine learning, specifically transformer architecture, have led to significant advancements in commercial domains. These powerful models have demonstrated superior capability to learn complex relationships and often generalize better to new data and problems. This paper presents a novel transformer-powered approach for enhancing prediction accuracy in multi-modal output scenarios, where sparse experimental data is supplemented with simulation data. The proposed approach integrates transformer-based architecture with a novel graph-based hyper-parameter optimization technique. The resulting system not only effectively reduces simulation bias, but also achieves superior prediction accuracy compared to the prior method. We demonstrate the efficacy of our approach on inertial confinement fusion experiments, where only 10 shots of real-world data are available, as well as synthetic versions of these experiments.
Authors: Jeongwhan Choi, Hyowon Wi, Jayoung Kim, Yehjin Shin, Kookjin Lee, Nathaniel Trask, Noseong Park
Abstract: Transformers, renowned for their self-attention mechanism, have achieved state-of-the-art performance across various tasks in natural language processing, computer vision, time-series modeling, etc. However, one of the challenges with deep Transformer models is the oversmoothing problem, where representations across layers converge to indistinguishable values, leading to significant performance degradation. We interpret the original self-attention as a simple graph filter and redesign it from a graph signal processing (GSP) perspective. We propose a graph-filter-based self-attention (GFSA) to learn a general yet effective one, whose complexity, however, is slightly larger than that of the original self-attention mechanism. We demonstrate that GFSA improves the performance of Transformers in various fields, including computer vision, natural language processing, graph regression, speech recognition, and code classification.
Authors: Lingfan Yu, Jinyang Li
Abstract: Large Language Models (LLMs) are wildly popular today and it is important to serve them efficiently. Existing LLM serving systems are stateless across requests. Consequently, when LLMs are used in the common setting of multi-turn conversations, a growing log of the conversation history must be processed alongside any request by the serving system at each turn, resulting in repeated processing. In this paper, we design Pensieve, a system optimized for multi-turn conversation LLM serving. Pensieve maintains the conversation state across requests by caching previously processed history to avoid duplicate processing. Pensieve's multi-tier caching strategy can utilize both GPU and CPU memory to efficiently store and retrieve cached data. Pensieve also generalizes the recent PagedAttention kernel to support attention between multiple input tokens with a GPU cache spread over non-contiguous memory. Our evaluation shows that Pensieve can achieve 13-58% more throughput compared to vLLM and TensorRT-LLM and significantly reduce latency.
Authors: Jakub Bia{\l}ek, Wojtek Kuberski, Nikolaos Perrakis, Albert Bifet
Abstract: Machine learning models often experience performance degradation post-deployment due to shifts in data distribution. It is challenging to assess model's performance accurately when labels are missing or delayed. Existing proxy methods, such as drift detection, fail to measure the effects of these shifts adequately. To address this, we introduce a new method, Probabilistic Adaptive Performance Estimation (PAPE), for evaluating classification models on unlabeled data that accurately quantifies the impact of covariate shift on model performance. It is model and data-type agnostic and works for various performance metrics. Crucially, PAPE operates independently of the original model, relying only on its predictions and probability estimates, and does not need any assumptions about the nature of the covariate shift, learning directly from data instead. We tested PAPE on tabular data using over 900 dataset-model combinations created from US census data, assessing its performance against multiple benchmarks. Overall, PAPE provided more accurate performance estimates than other evaluated methodologies.
Authors: Zhijie Zhong, Zhiwen Yu, Yiyuan Yang, Weizheng Wang, Kaixiang Yang
Abstract: Anomaly detection in time series analysis is a pivotal task, yet it poses the challenge of discerning normal and abnormal patterns in label-deficient scenarios. While prior studies have largely employed reconstruction-based approaches, which limits the models' representational capacities. Moreover, existing deep learning-based methods are not sufficiently lightweight. Addressing these issues, we present PatchAD, our novel, highly efficient multiscale patch-based MLP-Mixer architecture that utilizes contrastive learning for representation extraction and anomaly detection. With its four distinct MLP Mixers and innovative dual project constraint module, PatchAD mitigates potential model degradation and offers a lightweight solution, requiring only $3.2$MB. Its efficacy is demonstrated by state-of-the-art results across $9$ datasets sourced from different application scenarios, outperforming over $30$ comparative algorithms. PatchAD significantly improves the classical F1 score by $50.5\%$, the Aff-F1 score by $7.8\%$, and the AUC by $10.0\%$. The code is publicly available. \url{https://github.com/EmorZz1G/PatchAD}
Authors: Wei Ju, Siyu Yi, Yifan Wang, Qingqing Long, Junyu Luo, Zhiping Xiao, Ming Zhang
Abstract: Graph-structured data, prevalent in domains ranging from social networks to biochemical analysis, serve as the foundation for diverse real-world systems. While graph neural networks demonstrate proficiency in modeling this type of data, their success is often reliant on significant amounts of labeled data, posing a challenge in practical scenarios with limited annotation resources. To tackle this problem, tremendous efforts have been devoted to enhancing graph machine learning performance under low-resource settings by exploring various approaches to minimal supervision. In this paper, we introduce a novel concept of Data-Efficient Graph Learning (DEGL) as a research frontier, and present the first survey that summarizes the current progress of DEGL. We initiate by highlighting the challenges inherent in training models with large labeled data, paving the way for our exploration into DEGL. Next, we systematically review recent advances on this topic from several key aspects, including self-supervised graph learning, semi-supervised graph learning, and few-shot graph learning. Also, we state promising directions for future research, contributing to the evolution of graph machine learning.
Authors: Yingyi Chen, Qinghua Tao, Francesco Tonin, Johan A. K. Suykens
Abstract: While the great capability of Transformers significantly boosts prediction accuracy, it could also yield overconfident predictions and require calibrated uncertainty estimation, which can be commonly tackled by Gaussian processes (GPs). Existing works apply GPs with symmetric kernels under variational inference to the attention kernel; however, omitting the fact that attention kernels are in essence asymmetric. Moreover, the complexity of deriving the GP posteriors remains high for large-scale data. In this work, we propose Kernel-Eigen Pair Sparse Variational Gaussian Processes (KEP-SVGP) for building uncertainty-aware self-attention where the asymmetry of attention kernels is tackled by Kernel SVD (KSVD) and a reduced complexity is acquired. Through KEP-SVGP, i) the SVGP pair induced by the two sets of singular vectors from KSVD w.r.t. the attention kernel fully characterizes the asymmetry; ii) using only a small set of adjoint eigenfunctions from KSVD, the derivation of SVGP posteriors can be based on the inversion of a diagonal matrix containing singular values, contributing to a reduction in time complexity; iii) an evidence lower bound is derived so that variational parameters and network weights can be optimized with it. Experiments verify our excellent performances and efficiency on in-distribution, distribution-shift and out-of-distribution benchmarks.
Authors: Emanuel Sommer, Lisa Wimmer, Theodore Papamarkou, Ludwig Bothmann, Bernd Bischl, David R\"ugamer
Abstract: A major challenge in sample-based inference (SBI) for Bayesian neural networks is the size and structure of the networks' parameter space. Our work shows that successful SBI is possible by embracing the characteristic relationship between weight and function space, uncovering a systematic link between overparameterization and the difficulty of the sampling problem. Through extensive experiments, we establish practical guidelines for sampling and convergence diagnosis. As a result, we present a deep ensemble initialized approach as an effective solution with competitive performance and uncertainty quantification.
Authors: Akshay K. Jagadish, Julian Coda-Forno, Mirko Thalmann, Eric Schulz, Marcel Binz
Abstract: Ecological rationality refers to the notion that humans are rational agents adapted to their environment. However, testing this theory remains challenging due to two reasons: the difficulty in defining what tasks are ecologically valid and building rational models for these tasks. In this work, we demonstrate that large language models can generate cognitive tasks, specifically category learning tasks, that match the statistics of real-world tasks, thereby addressing the first challenge. We tackle the second challenge by deriving rational agents adapted to these tasks using the framework of meta-learning, leading to a class of models called ecologically rational meta-learned inference (ERMI). ERMI quantitatively explains human data better than seven other cognitive models in two different experiments. It additionally matches human behavior on a qualitative level: (1) it finds the same tasks difficult that humans find difficult, (2) it becomes more reliant on an exemplar-based strategy for assigning categories with learning, and (3) it generalizes to unseen stimuli in a human-like way. Furthermore, we show that ERMI's ecologically valid priors allow it to achieve state-of-the-art performance on the OpenML-CC18 classification benchmark.
Authors: Hao Chen, Jindong Wang, Lei Feng, Xiang Li, Yidong Wang, Xing Xie, Masashi Sugiyama, Rita Singh, Bhiksha Raj
Abstract: Weakly supervised learning generally faces challenges in applicability to various scenarios with diverse weak supervision and in scalability due to the complexity of existing algorithms, thereby hindering the practical deployment. This paper introduces a general framework for learning from weak supervision (GLWS) with a novel algorithm. Central to GLWS is an Expectation-Maximization (EM) formulation, adeptly accommodating various weak supervision sources, including instance partial labels, aggregate statistics, pairwise observations, and unlabeled data. We further present an advanced algorithm that significantly simplifies the EM computational demands using a Non-deterministic Finite Automaton (NFA) along with a forward-backward algorithm, which effectively reduces time complexity from quadratic or factorial often required in existing solutions to linear scale. The problem of learning from arbitrary weak supervision is therefore converted to the NFA modeling of them. GLWS not only enhances the scalability of machine learning models but also demonstrates superior performance and versatility across 11 weak supervision scenarios. We hope our work paves the way for further advancements and practical deployment in this field.
Authors: Haowei Lin, Baizhou Huang, Haotian Ye, Qinyu Chen, Zihao Wang, Sujian Li, Jianzhu Ma, Xiaojun Wan, James Zou, Yitao Liang
Abstract: The ever-growing ecosystem of LLMs has posed a challenge in selecting the most appropriate pre-trained model to fine-tune amidst a sea of options. Given constrained resources, fine-tuning all models and making selections afterward is unrealistic. In this work, we formulate this resource-constrained selection task into predicting fine-tuning performance and illustrate its natural connection with Scaling Law. Unlike pre-training, we find that the fine-tuning scaling curve includes not just the well-known "power phase" but also the previously unobserved "pre-power phase". We also explain why existing Scaling Law fails to capture this phase transition phenomenon both theoretically and empirically. To address this, we introduce the concept of "pre-learned data size" into our Rectified Scaling Law, which overcomes theoretical limitations and fits experimental results much better. By leveraging our law, we propose a novel LLM selection algorithm that selects the near-optimal model with hundreds of times less resource consumption, while other methods may provide negatively correlated selection. The project page is available at rectified-scaling-law.github.io.
Authors: Naoki Sato, Hideaki Iiduka
Abstract: For nonconvex objective functions, including deep neural networks, stochastic gradient descent (SGD) with momentum has fast convergence and excellent generalizability, but a theoretical explanation for this is lacking. In contrast to previous studies that defined the stochastic noise that occurs during optimization as the variance of the stochastic gradient, we define it as the gap between the search direction of the optimizer and the steepest descent direction and show that its level dominates generalizability of the model. We also show that the stochastic noise in SGD with momentum smoothes the objective function, the degree of which is determined by the learning rate, the batch size, the momentum factor, the variance of the stochastic gradient, and the upper bound of the gradient norm. By numerically deriving the stochastic noise level in SGD and SGD with momentum, we provide theoretical findings that help explain the training dynamics of SGD with momentum, which were not explained by previous studies on convergence and stability. We also provide experimental results supporting our assertion that model generalizability depends on the stochastic noise level.
Authors: Arsalan Sharifnassab, Saber Salehkaleybar, Richard Sutton
Abstract: This paper addresses the challenge of optimizing meta-parameters (i.e., hyperparameters) in machine learning algorithms, a critical factor influencing training efficiency and model performance. Moving away from the computationally expensive traditional meta-parameter search methods, we introduce MetaOptimize framework that dynamically adjusts meta-parameters, particularly step sizes (also known as learning rates), during training. More specifically, MetaOptimize can wrap around any first-order optimization algorithm, tuning step sizes on the fly to minimize a specific form of regret that accounts for long-term effect of step sizes on training, through a discounted sum of future losses. We also introduce low complexity variants of MetaOptimize that, in conjunction with its adaptability to multiple optimization algorithms, demonstrate performance competitive to those of best hand-crafted learning rate schedules across various machine learning applications.
Authors: Edwin V. Bonilla, Pantelis Elinas, He Zhao, Maurizio Filippone, Vassili Kitsios, Terry O'Kane
Abstract: Estimating the structure of a Bayesian network, in the form of a directed acyclic graph (DAG), from observational data is a statistically and computationally hard problem with essential applications in areas such as causal discovery. Bayesian approaches are a promising direction for solving this task, as they allow for uncertainty quantification and deal with well-known identifiability issues. From a probabilistic inference perspective, the main challenges are (i) representing distributions over graphs that satisfy the DAG constraint and (ii) estimating a posterior over the underlying combinatorial space. We propose an approach that addresses these challenges by formulating a joint distribution on an augmented space of DAGs and permutations. We carry out posterior estimation via variational inference, where we exploit continuous relaxations of discrete distributions. We show that our approach performs competitively when compared with a wide range of Bayesian and non-Bayesian benchmarks on a range of synthetic and real datasets.
Authors: Enneng Yang, Li Shen, Zhenyi Wang, Guibing Guo, Xiaojun Chen, Xingwei Wang, Dacheng Tao
Abstract: Multi-task learning (MTL) compresses the information from multiple tasks into a unified backbone to improve computational efficiency and generalization. Recent work directly merges multiple independently trained models to perform MTL instead of collecting their raw data for joint training, greatly expanding the application scenarios of MTL. However, by visualizing the representation distribution of existing model merging schemes, we find that the merged model often suffers from the dilemma of representation bias. That is, there is a significant discrepancy in the representation distribution between the merged and individual models, resulting in poor performance of merged MTL. In this paper, we propose a representation surgery solution called "Surgery" to reduce representation bias in the merged model. Specifically, Surgery is a lightweight task-specific module that takes the representation of the merged model as input and attempts to output the biases contained in the representation from the merged model. We then designed an unsupervised optimization objective that updates the Surgery module by minimizing the distance between the merged model's representation and the individual model's representation. Extensive experiments demonstrate significant MTL performance improvements when our Surgery module is applied to state-of-the-art (SOTA) model merging schemes.
Authors: Vinitra Swamy, Syrielle Montariol, Julian Blackwell, Jibril Frej, Martin Jaggi, Tanja K\"aser
Abstract: Interpretability for neural networks is a trade-off between three key requirements: 1) faithfulness of the explanation (i.e., how perfectly it explains the prediction), 2) understandability of the explanation by humans, and 3) model performance. Most existing methods compromise one or more of these requirements; e.g., post-hoc approaches provide limited faithfulness, automatically identified feature masks compromise understandability, and intrinsically interpretable methods such as decision trees limit model performance. These shortcomings are unacceptable for sensitive applications such as education and healthcare, which require trustworthy explanations, actionable interpretations, and accurate predictions. In this work, we present InterpretCC (interpretable conditional computation), a family of interpretable-by-design neural networks that guarantee human-centric interpretability, while maintaining comparable performance to state-of-the-art models by adaptively and sparsely activating features before prediction. We extend this idea into an interpretable, global mixture-of-experts (MoE) model that allows humans to specify topics of interest, discretely separates the feature space for each data point into topical subnetworks, and adaptively and sparsely activates these topical subnetworks for prediction. We apply variations of the InterpretCC architecture for text, time series and tabular data across several real-world benchmarks, demonstrating comparable performance with non-interpretable baselines, outperforming interpretable-by-design baselines, and showing higher actionability and usefulness according to a user study.
Authors: Zhanpeng Zhou, Zijun Chen, Yilan Chen, Bo Zhang, Junchi Yan
Abstract: The pretraining-finetuning paradigm has become the prevailing trend in modern deep learning. In this work, we discover an intriguing linear phenomenon in models that are initialized from a common pretrained checkpoint and finetuned on different tasks, termed as Cross-Task Linearity (CTL). Specifically, we show that if we linearly interpolate the weights of two finetuned models, the features in the weight-interpolated model are often approximately equal to the linear interpolation of features in two finetuned models at each layer. We provide comprehensive empirical evidence supporting that CTL consistently occurs for finetuned models that start from the same pretrained checkpoint. We conjecture that in the pretraining-finetuning paradigm, neural networks approximately function as linear maps, mapping from the parameter space to the feature space. Based on this viewpoint, our study unveils novel insights into explaining model merging/editing, particularly by translating operations from the parameter space to the feature space. Furthermore, we delve deeper into the root cause for the emergence of CTL, highlighting the role of pretraining.
Authors: Tsunehiko Tanaka, Kenshi Abe, Kaito Ariu, Tetsuro Morimura, Edgar Simo-Serra
Abstract: Traditional approaches in offline reinforcement learning aim to learn the optimal policy that maximizes the cumulative reward, also known as return. However, as applications broaden, it becomes increasingly crucial to train agents that not only maximize the returns, but align the actual return with a specified target return, giving control over the agent's performance. Decision Transformer (DT) optimizes a policy that generates actions conditioned on the target return through supervised learning and is equipped with a mechanism to control the agent using the target return. However, the action generation is hardly influenced by the target return because DT's self-attention allocates scarce attention scores to the return tokens. In this paper, we propose Return-Aligned Decision Transformer (RADT), designed to effectively align the actual return with the target return. RADT utilizes features extracted by paying attention solely to the return, enabling the action generation to consistently depend on the target return. Extensive experiments show that RADT reduces the discrepancies between the actual return and the target return of DT-based methods.
Authors: Deqian Kong, Dehong Xu, Minglu Zhao, Bo Pang, Jianwen Xie, Andrew Lizarraga, Yuhao Huang, Sirui Xie, Ying Nian Wu
Abstract: In tasks aiming for long-term returns, planning becomes essential. We study generative modeling for planning with datasets repurposed from offline reinforcement learning. Specifically, we identify temporal consistency in the absence of step-wise rewards as one key technical challenge. We introduce the Latent Plan Transformer (LPT), a novel model that leverages a latent space to connect a Transformer-based trajectory generator and the final return. LPT can be learned with maximum likelihood estimation on trajectory-return pairs. In learning, posterior sampling of the latent variable naturally integrates sub-trajectories to form a consistent abstraction despite the finite context. At test time, the latent variable is inferred from an expected return before policy execution, realizing the idea of planning as inference. Our experiments demonstrate that LPT can discover improved decisions from suboptimal trajectories, achieving competitive performance across several benchmarks, including Gym-Mujoco, Franka Kitchen, Maze2D, and Connect Four. It exhibits capabilities in nuanced credit assignments, trajectory stitching, and adaptation to environmental contingencies. These results validate that latent variable inference can be a strong alternative to step-wise reward prompting.
Authors: Biyonka Liang, Lily Xu, Aparna Taneja, Milind Tambe, Lucas Janson
Abstract: Public health programs often provide interventions to encourage beneficiary adherence,and effectively allocating interventions is vital for producing the greatest overall health outcomes. Such resource allocation problems are often modeled as restless multi-armed bandits (RMABs) with unknown underlying transition dynamics, hence requiring online reinforcement learning (RL). We present Bayesian Learning for Contextual RMABs (BCoR), an online RL approach for RMABs that novelly combines techniques in Bayesian modeling with Thompson sampling to flexibly model the complex RMAB settings present in public health program adherence problems, such as context and non-stationarity. BCoR's key strength is the ability to leverage shared information within and between arms to learn the unknown RMAB transition dynamics quickly in intervention-scarce settings with relatively short time horizons, which is common in public health applications. Empirically, BCoR achieves substantially higher finite-sample performance over a range of experimental settings, including an example based on real-world adherence data that was developed in collaboration with ARMMAN, an NGO in India which runs a large-scale maternal health program, showcasing BCoR practical utility and potential for real-world deployment.
Authors: Itay Lavie, Guy Gur-Ari, Zohar Ringel
Abstract: We study inductive bias in Transformers in the infinitely over-parameterized Gaussian process limit and argue transformers tend to be biased towards more permutation symmetric functions in sequence space. We show that the representation theory of the symmetric group can be used to give quantitative analytical predictions when the dataset is symmetric to permutations between tokens. We present a simplified transformer block and solve the model at the limit, including accurate predictions for the learning curves and network outputs. We show that in common setups, one can derive tight bounds in the form of a scaling law for the learnability as a function of the context length. Finally, we argue WikiText dataset, does indeed possess a degree of permutation symmetry.
Authors: Lior Cohen, Kaixin Wang, Bingyi Kang, Shie Mannor
Abstract: Motivated by the success of Transformers when applied to sequences of discrete symbols, token-based world models (TBWMs) were recently proposed as sample-efficient methods. In TBWMs, the world model consumes agent experience as a language-like sequence of tokens, where each observation constitutes a sub-sequence. However, during imagination, the sequential token-by-token generation of next observations results in a severe bottleneck, leading to long training times, poor GPU utilization, and limited representations. To resolve this bottleneck, we devise a novel Parallel Observation Prediction (POP) mechanism. POP augments a Retentive Network (RetNet) with a novel forward mode tailored to our reinforcement learning setting. We incorporate POP in a novel TBWM agent named REM (Retentive Environment Model), showcasing a 15.4x faster imagination compared to prior TBWMs. REM attains superhuman performance on 12 out of 26 games of the Atari 100K benchmark, while training in less than 12 hours. Our code is available at \url{https://github.com/leor-c/REM}.
Authors: Orson Mengara
Abstract: Diffusion models are state-of-the-art deep learning generative models that are trained on the principle of learning forward and backward diffusion processes via the progressive addition of noise and denoising. In this paper, we aim to fool audio-based DNN models, such as those from the Hugging Face framework, primarily those that focus on audio, in particular transformer-based artificial intelligence models, which are powerful machine learning models that save time and achieve results faster and more efficiently. We demonstrate the feasibility of backdoor attacks (called `BacKBayDiffMod`) on audio transformers derived from Hugging Face, a popular framework in the world of artificial intelligence research. The backdoor attack developed in this paper is based on poisoning model training data uniquely by incorporating backdoor diffusion sampling and a Bayesian approach to the distribution of poisoned data.
Authors: Marc Bartholet, Taehyeon Kim, Ami Beuret, Se-Young Yun, Joachim M. Buhmann
Abstract: Federated Learning (FL) faces significant challenges with domain shifts in heterogeneous data, degrading performance. Traditional domain generalization aims to learn domain-invariant features, but the federated nature of model averaging often limits this due to its linear aggregation of local learning. To address this, we propose a robust framework, coined as hypernetwork-based Federated Fusion (hFedF), using hypernetworks for non-linear aggregation, facilitating generalization to unseen domains. Our method employs client-specific embeddings and gradient alignment techniques to manage domain generalization effectively. Evaluated in both zero-shot and few-shot settings, hFedF demonstrates superior performance in handling domain shifts. Comprehensive comparisons on PACS, Office-Home, and VLCS datasets show that hFedF consistently achieves the highest in-domain and out-of-domain accuracy with reliable predictions. Our study contributes significantly to the under-explored field of Federated Domain Generalization (FDG), setting a new benchmark for performance in this area.
Authors: Billy J. Franks, Christopher Morris, Ameya Velingker, Floris Geerts
Abstract: The Weisfeiler-Leman algorithm ($1$-WL) is a well-studied heuristic for the graph isomorphism problem. Recently, the algorithm has played a prominent role in understanding the expressive power of message-passing graph neural networks (MPNNs) and being effective as a graph kernel. Despite its success, $1$-WL faces challenges in distinguishing non-isomorphic graphs, leading to the development of more expressive MPNN and kernel architectures. However, the relationship between enhanced expressivity and improved generalization performance remains unclear. Here, we show that an architecture's expressivity offers limited insights into its generalization performance when viewed through graph isomorphism. Moreover, we focus on augmenting $1$-WL and MPNNs with subgraph information and employ classical margin theory to investigate the conditions under which an architecture's increased expressivity aligns with improved generalization performance. In addition, we show that gradient flow pushes the MPNN's weights toward the maximum margin solution. Further, we introduce variations of expressive $1$-WL-based kernel and MPNN architectures with provable generalization properties. Our empirical study confirms the validity of our theoretical findings.
Authors: Guy Bar-Shalom, Beatrice Bevilacqua, Haggai Maron
Abstract: In the realm of Graph Neural Networks (GNNs), two exciting research directions have recently emerged: Subgraph GNNs and Graph Transformers. In this paper, we propose an architecture that integrates both approaches, dubbed Subgraphormer, which combines the enhanced expressive power, message-passing mechanisms, and aggregation schemes from Subgraph GNNs with attention and positional encodings, arguably the most important components in Graph Transformers. Our method is based on an intriguing new connection we reveal between Subgraph GNNs and product graphs, suggesting that Subgraph GNNs can be formulated as Message Passing Neural Networks (MPNNs) operating on a product of the graph with itself. We use this formulation to design our architecture: first, we devise an attention mechanism based on the connectivity of the product graph. Following this, we propose a novel and efficient positional encoding scheme for Subgraph GNNs, which we derive as a positional encoding for the product graph. Our experimental results demonstrate significant performance improvements over both Subgraph GNNs and Graph Transformers on a wide range of datasets.
Authors: Daniel D. Johnson, Daniel Tarlow, David Duvenaud, Chris J. Maddison
Abstract: Identifying how much a model ${\widehat{p}}_{\theta}(Y|X)$ knows about the stochastic real-world process $p(Y|X)$ it was trained on is important to ensure it avoids producing incorrect or "hallucinated" answers or taking unsafe actions. But this is difficult for generative models because probabilistic predictions do not distinguish between per-response noise (aleatoric uncertainty) and lack of knowledge about the process (epistemic uncertainty), and existing epistemic uncertainty quantification techniques tend to be overconfident when the model underfits. We propose a general strategy for teaching a model to both approximate $p(Y|X)$ and also estimate the remaining gaps between ${\widehat{p}}_{\theta}(Y|X)$ and $p(Y|X)$: train it to predict pairs of independent responses drawn from the true conditional distribution, allow it to "cheat" by observing one response while predicting the other, then measure how much it cheats. Remarkably, we prove that being good at cheating (i.e. cheating whenever it improves your prediction) is equivalent to being second-order calibrated, a principled extension of ordinary calibration that allows us to construct provably-correct frequentist confidence intervals for $p(Y|X)$ and detect incorrect responses with high probability. We demonstrate empirically that our approach accurately estimates how much models don't know across ambiguous image classification, (synthetic) language modeling, and partially-observable navigation tasks, outperforming existing techniques.
Authors: Tristan Deleu, Padideh Nouri, Nikolay Malkin, Doina Precup, Yoshua Bengio
Abstract: We consider the problem of sampling from a discrete and structured distribution as a sequential decision problem, where the objective is to find a stochastic policy such that objects are sampled at the end of this sequential process proportionally to some predefined reward. While we could use maximum entropy Reinforcement Learning (MaxEnt RL) to solve this problem for some distributions, it has been shown that in general, the distribution over states induced by the optimal policy may be biased in cases where there are multiple ways to generate the same object. To address this issue, Generative Flow Networks (GFlowNets) learn a stochastic policy that samples objects proportionally to their reward by approximately enforcing a conservation of flows across the whole Markov Decision Process (MDP). In this paper, we extend recent methods correcting the reward in order to guarantee that the marginal distribution induced by the optimal MaxEnt RL policy is proportional to the original reward, regardless of the structure of the underlying MDP. We also prove that some flow-matching objectives found in the GFlowNet literature are in fact equivalent to well-established MaxEnt RL algorithms with a corrected reward. Finally, we study empirically the performance of multiple MaxEnt RL and GFlowNet algorithms on multiple problems involving sampling from discrete distributions.
Authors: Tim Tsz-Kit Lau, Han Liu, Mladen Kolar
Abstract: The choice of batch sizes in minibatch stochastic gradient optimizers is critical in large-scale model training for both optimization and generalization performance. Although large-batch training is arguably the dominant training paradigm for large-scale deep learning due to hardware advances, the generalization performance of the model deteriorates compared to small-batch training, leading to the so-called "generalization gap" phenomenon. To mitigate this, we investigate adaptive batch size strategies derived from adaptive sampling methods, originally developed only for stochastic gradient descent. Given the significant interplay between learning rates and batch sizes, and considering the prevalence of adaptive gradient methods in deep learning, we emphasize the need for adaptive batch size strategies in these contexts. We introduce AdAdaGrad and its scalar variant AdAdaGradNorm, which progressively increase batch sizes during training, while model updates are performed using AdaGrad and AdaGradNorm. We prove that AdAdaGradNorm converges with high probability at a rate of $\mathscr{O}(1/K)$ to find a first-order stationary point of smooth nonconvex functions within $K$ iterations. AdAdaGrad also demonstrates similar convergence properties when integrated with a novel coordinate-wise variant of our adaptive batch size strategies. We corroborate our theoretical claims by performing image classification experiments, highlighting the merits of the proposed schemes in terms of both training efficiency and model generalization. Our work unveils the potential of adaptive batch size strategies for adaptive gradient optimizers in large-scale model training.
Authors: Yihua Zhang, Pingzhi Li, Junyuan Hong, Jiaxiang Li, Yimeng Zhang, Wenqing Zheng, Pin-Yu Chen, Jason D. Lee, Wotao Yin, Mingyi Hong, Zhangyang Wang, Sijia Liu, Tianlong Chen
Abstract: In the evolving landscape of natural language processing (NLP), fine-tuning pre-trained Large Language Models (LLMs) with first-order (FO) optimizers like SGD and Adam has become standard. Yet, as LLMs grow {in size}, the substantial memory overhead from back-propagation (BP) for FO gradient computation presents a significant challenge. Addressing this issue is crucial, especially for applications like on-device training where memory efficiency is paramount. This paper proposes a shift towards BP-free, zeroth-order (ZO) optimization as a solution for reducing memory costs during LLM fine-tuning, building on the initial concept introduced by MeZO. Unlike traditional ZO-SGD methods, our work expands the exploration to a wider array of ZO optimization techniques, through a comprehensive, first-of-its-kind benchmarking study across five LLM families (Roberta, OPT, LLaMA, Vicuna, Mistral), three task complexities, and five fine-tuning schemes. Our study unveils previously overlooked optimization principles, highlighting the importance of task alignment, the role of the forward gradient method, and the balance between algorithm complexity and fine-tuning performance. We further introduce novel enhancements to ZO optimization, including block-wise descent, hybrid training, and gradient sparsity. Our study offers a promising direction for achieving further memory-efficient LLM fine-tuning. Codes to reproduce all our experiments are at https://github.com/ZO-Bench/ZO-LLM .
Authors: Liam Collins, Advait Parulekar, Aryan Mokhtari, Sujay Sanghavi, Sanjay Shakkottai
Abstract: A striking property of transformers is their ability to perform in-context learning (ICL), a machine learning framework in which the learner is presented with a novel context during inference implicitly through some data, and tasked with making a prediction in that context. As such, that learner must adapt to the context without additional training. We explore the role of softmax attention in an ICL setting where each context encodes a regression task. We show that an attention unit learns a window that it uses to implement a nearest-neighbors predictor adapted to the landscape of the pretraining tasks. Specifically, we show that this window widens with decreasing Lipschitzness and increasing label noise in the pretraining tasks. We also show that on low-rank, linear problems, the attention unit learns to project onto the appropriate subspace before inference. Further, we show that this adaptivity relies crucially on the softmax activation and thus cannot be replicated by the linear activation often studied in prior theoretical analyses.
Authors: Uijeong Jang, Jason D. Lee, Ernest K. Ryu
Abstract: Low-rank adaptation (LoRA) has become the standard approach for parameter-efficient fine-tuning of large language models (LLM), but our theoretical understanding of LoRA has been limited. In this work, we theoretically analyze LoRA fine-tuning in the neural tangent kernel (NTK) regime with $N$ data points, showing: (i) full fine-tuning (without LoRA) admits a low-rank solution of rank $r\lesssim \sqrt{N}$; (ii) using LoRA with rank $r\gtrsim \sqrt{N}$ eliminates spurious local minima, allowing gradient descent to find the low-rank solutions; (iii) the low-rank solution found using LoRA generalizes well.
Authors: Hezhe Qiao, Qingsong Wen, Xiaoli Li, Ee-Peng Lim, Guansong Pang
Abstract: This work considers a practical semi-supervised graph anomaly detection (GAD) scenario, where part of the nodes in a graph are known to be normal, contrasting to the extensively explored unsupervised setting with a fully unlabeled graph. We reveal that having access to the normal nodes, even just a small percentage of normal nodes, helps enhance the detection performance of existing unsupervised GAD methods when they are adapted to the semi-supervised setting. However, their utilization of these normal nodes is limited. In this paper, we propose a novel Generative GAD approach (namely GGAD) for the semi-supervised scenario to better exploit the normal nodes. The key idea is to generate pseudo anomaly nodes, referred to as 'outlier nodes', for providing effective negative node samples in training a discriminative one-class classifier. The main challenge here lies in the lack of ground truth information about real anomaly nodes. To address this challenge, GGAD is designed to leverage two important priors about the anomaly nodes -- asymmetric local affinity and egocentric closeness -- to generate reliable outlier nodes that assimilate anomaly nodes in both graph structure and feature representations. Comprehensive experiments on six real-world GAD datasets are performed to establish a benchmark for semi-supervised GAD and show that GGAD substantially outperforms state-of-the-art unsupervised and semi-supervised GAD methods with varying numbers of training normal nodes. Code will be made available at https://github.com/mala-lab/GGAD.
Authors: Kai Wang, Zhaopan Xu, Yukun Zhou, Zelin Zang, Trevor Darrell, Zhuang Liu, Yang You
Abstract: Diffusion models have achieved remarkable success in image and video generation. In this work, we demonstrate that diffusion models can also \textit{generate high-performing neural network parameters}. Our approach is simple, utilizing an autoencoder and a standard latent diffusion model. The autoencoder extracts latent representations of a subset of the trained network parameters. A diffusion model is then trained to synthesize these latent parameter representations from random noise. It then generates new representations that are passed through the autoencoder's decoder, whose outputs are ready to use as new subsets of network parameters. Across various architectures and datasets, our diffusion process consistently generates models of comparable or improved performance over trained networks, with minimal additional cost. Notably, we empirically find that the generated models are not memorizing the trained networks. Our results encourage more exploration on the versatile use of diffusion models.
Authors: Siyuan Guo, Cheng Deng, Ying Wen, Hechang Chen, Yi Chang, Jun Wang
Abstract: In this work, we investigate the potential of large language models (LLMs) based agents to automate data science tasks, with the goal of comprehending task requirements, then building and training the best-fit machine learning models. Despite their widespread success, existing LLM agents are hindered by generating unreasonable experiment plans within this scenario. To this end, we present DS-Agent, a novel automatic framework that harnesses LLM agent and case-based reasoning (CBR). In the development stage, DS-Agent follows the CBR framework to structure an automatic iteration pipeline, which can flexibly capitalize on the expert knowledge from Kaggle, and facilitate consistent performance improvement through the feedback mechanism. Moreover, DS-Agent implements a low-resource deployment stage with a simplified CBR paradigm to adapt past successful solutions from the development stage for direct code generation, significantly reducing the demand on foundational capabilities of LLMs. Empirically, DS-Agent with GPT-4 achieves 100\% success rate in the development stage, while attaining 36\% improvement on average one pass rate across alternative LLMs in the deployment stage. In both stages, DS-Agent achieves the best rank in performance, costing \$1.60 and \$0.13 per run with GPT-4, respectively. Our data and code are open-sourced at https://github.com/guosyjlu/DS-Agent.
Authors: Keying Kuang, Frances Dean, Jack B. Jedlicki, David Ouyang, Anthony Philippakis, David Sontag, Ahmed M. Alaa
Abstract: A digital twin is a virtual replica of a real-world physical phenomena that uses mathematical modeling to characterize and simulate its defining features. By constructing digital twins for disease processes, we can perform in-silico simulations that mimic patients' health conditions and counterfactual outcomes under hypothetical interventions in a virtual setting. This eliminates the need for invasive procedures or uncertain treatment decisions. In this paper, we propose a method to identify digital twin model parameters using only noninvasive patient health data. We approach the digital twin modeling as a composite inverse problem, and observe that its structure resembles pretraining and finetuning in self-supervised learning (SSL). Leveraging this, we introduce a physics-informed SSL algorithm that initially pretrains a neural network on the pretext task of learning a differentiable simulator of a physiological process. Subsequently, the model is trained to reconstruct physiological measurements from noninvasive modalities while being constrained by the physical equations learned in pretraining. We apply our method to identify digital twins of cardiac hemodynamics using noninvasive echocardiogram videos, and demonstrate its utility in unsupervised disease detection and in-silico clinical trials.
Authors: Awni Altabaa, Zhuoran Yang
Abstract: In a sequential decision-making problem, the information structure is the description of how events in the system occurring at different points in time affect each other. Classical models of reinforcement learning (e.g., MDPs, POMDPs) assume a simple and highly regular information structure, while more general models like predictive state representations do not explicitly model the information structure. By contrast, real-world sequential decision-making problems typically involve a complex and time-varying interdependence of system variables, requiring a rich and flexible representation of information structure. In this paper, we formalize a novel reinforcement learning model which explicitly represents the information structure. We then use this model to carry out an information-structural analysis of the statistical hardness of general sequential decision-making problems, obtaining a characterization via a graph-theoretic quantity of the DAG representation of the information structure. We prove an upper bound on the sample complexity of learning a general sequential decision-making problem in terms of its information structure by exhibiting an algorithm achieving the upper bound. This recovers known tractability results and gives a novel perspective on reinforcement learning in general sequential decision-making problems, providing a systematic way of identifying new tractable classes of problems.
Authors: Spencer Hutchinson, Tianyi Chen, Mahnoosh Alizadeh
Abstract: We study the problem of online convex optimization (OCO) under unknown linear constraints that are either static, or stochastically time-varying. For this problem, we introduce an algorithm that we term Optimistically Safe OCO (OSOCO) and show that it enjoys $\tilde{\mathcal{O}}(\sqrt{T})$ regret and no constraint violation. In the case of static linear constraints, this improves on the previous best known $\tilde{\mathcal{O}}(T^{2/3})$ regret with only slightly stronger assumptions. In the case of stochastic time-varying constraints, our work supplements existing results that show $\mathcal{O}(\sqrt{T})$ regret and $\mathcal{O}(\sqrt{T})$ cumulative violation under more general convex constraints albeit a less general feedback model. In addition to our theoretical guarantees, we also give numerical results comparing the performance of OSOCO to existing algorithms.
Authors: Pierre Boyeau, Anastasios N. Angelopoulos, Nir Yosef, Jitendra Malik, Michael I. Jordan
Abstract: The evaluation of machine learning models using human-labeled validation data can be expensive and time-consuming. AI-labeled synthetic data can be used to decrease the number of human annotations required for this purpose in a process called autoevaluation. We suggest efficient and statistically principled algorithms for this purpose that improve sample efficiency while remaining unbiased. These algorithms increase the effective human-labeled sample size by up to 50% on experiments with GPT-4.
Authors: Yunrui Li, Hao Xu, Pengyu Hong
Abstract: Nuclear magnetic resonance (NMR) spectroscopy is crucial across diverse scientific fields, revealing detailed structural information, electronic properties, and molecular dynamic insights. Accurate prediction of NMR peaks in a spectrum from molecular structures allows chemists to effectively evaluate candidate structures by comparing predictions with experimental shifts in an NMR spectra. This process facilitates peak assignments, thereby aiding in verifying molecular structures or identifying discrepancies. Although significant progress has been made in predicting 1D NMR with Machine Learning (ML) approaches, 2D NMR prediction remains a challenge due to the lack of an annotated 2D NMR training dataset. To address this gap, we propose an Iterative Unsupervised Learning (IUL) approach to train a machine learning model for predicting atomic 2D NMR cross peaks and annotating peaks in experimental 2D NMR spectra. Initially, the model undergoes a Multi-Task pre-Training (MTT) phase using a set of annotated 1D 1H and 13C NMR spectra. Then, the model is iteratively improved through a fine-tuning process with IUL, alternating between using the model to annotate the unlabeled 2D NMR data and refining the model using the newly generated annotations. Using the proposed approach, we trained our model on 19,000 Heteronuclear Single Quantum Coherence (HSQC) spectra, tested it on 500 HSQC spectra with expert annotations, and further compared it with two traditional methods (ChemDraw and Mestrenova) on another expert-annotated HSQC dataset. For HSQC cross peak prediction, our model achieves MAE of 2.035 ppm and 0.163 ppm for 13C shifts and 1H shifts on the test dataset, respectively, and outperforms the conventional tools. This performance demonstrates not only the model's capability in accurately predicting chemical shifts, but also its effectiveness in peak assignments for experimental HSQC spectra.
Authors: Yilei Chen, Aldo Pacchiano, Ioannis Ch. Paschalidis
Abstract: We study the multiple-policy evaluation problem where we are given a set of $K$ policies and the goal is to evaluate their performance (expected total reward over a fixed horizon) to an accuracy $\epsilon$ with probability at least $1-\delta$. We propose an algorithm named $\mathrm{CAESAR}$ for this problem. Our approach is based on computing an approximate optimal offline sampling distribution and using the data sampled from it to perform the simultaneous estimation of the policy values. $\mathrm{CAESAR}$ has two phases. In the first we produce coarse estimates of the visitation distributions of the target policies at a low order sample complexity rate that scales with $\tilde{O}(\frac{1}{\epsilon})$. In the second phase, we approximate the optimal offline sampling distribution and compute the importance weighting ratios for all target policies by minimizing a step-wise quadratic loss function inspired by the DualDICE \cite{nachum2019dualdice} objective. Up to low order and logarithmic terms $\mathrm{CAESAR}$ achieves a sample complexity $\tilde{O}\left(\frac{H^4}{\epsilon^2}\sum_{h=1}^H\max_{k\in[K]}\sum_{s,a}\frac{(d_h^{\pi^k}(s,a))^2}{\mu^*_h(s,a)}\right)$, where $d^{\pi}$ is the visitation distribution of policy $\pi$, $\mu^*$ is the optimal sampling distribution, and $H$ is the horizon.
Authors: Yuezhou Hu, Kang Zhao, Weiyu Huang, Jianfei Chen, Jun Zhu
Abstract: Training large transformers is slow, but recent innovations on GPU architecture give us an advantage. NVIDIA Ampere GPUs can execute a fine-grained 2:4 sparse matrix multiplication twice as fast as its dense equivalent. In the light of this property, we comprehensively investigate the feasibility of accelerating feed-forward networks (FFNs) of transformers in pre-training. First, we define a ``flip rate'' to monitor the stability of a 2:4 training process. Utilizing this metric, we propose three techniques to preserve accuracy: to modify the sparse-refined straight-through estimator by applying the masked decay term on gradients, to determine a feasible decay factor in warm-up stage, and to enhance the model's quality by a dense fine-tuning procedure near the end of pre-training. Besides, we devise two techniques to practically accelerate training: to calculate transposable 2:4 masks by convolution, and to accelerate gated activation functions by reducing GPU L2 cache miss. Experiments show that our 2:4 sparse training algorithm achieves similar convergence to dense training algorithms on several transformer pre-training tasks, while actual acceleration can be observed on different shapes of transformer block apparently. Our toolkit is available at https://github.com/huyz2023/2by4-pretrain.
Authors: Fanxu Meng, Zhaohui Wang, Muhan Zhang
Abstract: To parameter-efficiently fine-tune (PEFT) large language models (LLMs), the low-rank adaptation (LoRA) method approximates the model changes $\Delta W \in \mathbb{R}^{m \times n}$ through the product of two matrices $A \in \mathbb{R}^{m \times r}$ and $B \in \mathbb{R}^{r \times n}$, where $r \ll \min(m, n)$, $A$ is initialized with Gaussian noise, and $B$ with zeros. LoRA freezes the original model $W$ and updates the "Noise & Zero" adapter, which may lead to slow convergence. To overcome this limitation, we introduce Principal Singular values and Singular vectors Adaptation (PiSSA). PiSSA shares the same architecture as LoRA, but initializes the adaptor matrices $A$ and $B$ with the principal components of the original matrix $W$, and put the remaining components into a residual matrix $W^{res} \in \mathbb{R}^{m \times n}$ which is frozen during fine-tuning. Compared to LoRA, PiSSA updates the principal components while freezing the "residual" parts, allowing faster convergence and enhanced performance. Comparative experiments of PiSSA and LoRA across 12 different models, ranging from 184M to 70B, encompassing 5 NLG and 8 NLU tasks, reveal that PiSSA consistently outperforms LoRA under identical experimental setups. On the GSM8K benchmark, Mistral-7B fine-tuned with PiSSA achieves an accuracy of 72.86%, surpassing LoRA's 67.7% by 5.16%. Due to the same architecture, PiSSA is also compatible with quantization to further reduce the memory requirement of fine-tuning. Compared to QLoRA, QPiSSA (PiSSA with 4-bit quantization) exhibits smaller quantization errors in the initial stages. Fine-tuning LLaMA-3-70B on GSM8K, QPiSSA attains an accuracy of 86.05%, exceeding the performances of QLoRA at 81.73%. Leveraging a fast SVD technique, PiSSA can be initialized in only a few seconds, presenting a negligible cost for transitioning from LoRA to PiSSA.
Authors: Xize Liang, Chao Chen, Shuang Qiu, Jie Wang, Yue Wu, Zhihang Fu, Zhihao Shi, Feng Wu, Jieping Ye
Abstract: Preference alignment is pivotal for empowering large language models (LLMs) to generate helpful and harmless responses. However, the performance of preference alignment is highly sensitive to the prevalent noise in the preference data. Recent efforts for this problem either marginally alleviate the impact of noise without the ability to actually reduce its presence, or rely on costly teacher LLMs prone to reward misgeneralization. To address these challenges, we propose the RObust Preference Optimization (ROPO) framework, an iterative alignment approach that integrates noise-tolerance and filtering of noisy samples without the aid of external models. Specifically, ROPO iteratively solves a constrained optimization problem, where we dynamically assign a quality-aware weight for each sample and constrain the sum of the weights to the number of samples we intend to retain. For noise-tolerant training and effective noise identification, we derive a robust loss by suppressing the gradients of samples with high uncertainty. We demonstrate both empirically and theoretically that the derived loss is critical for distinguishing noisy samples from clean ones. Furthermore, inspired by our derived loss, we propose a robustness-guided rejection sampling technique to compensate for the potential important information in discarded queries. Experiments on three widely-used datasets with Mistral-7B and Llama-2-7B demonstrate that ROPO significantly outperforms existing preference alignment methods, with its superiority growing as the noise rate increases.
Authors: Peyman Gholami, Hulya Seferoglu
Abstract: This paper focuses on reducing the communication cost of federated learning by exploring generalization bounds and representation learning. We first characterize a tighter generalization bound for one-round federated learning based on local clients' generalizations and heterogeneity of data distribution (non-iid scenario). We also characterize a generalization bound in R-round federated learning and its relation to the number of local updates (local stochastic gradient descents (SGDs)). Then, based on our generalization bound analysis and our representation learning interpretation of this analysis, we show for the first time that less frequent aggregations, hence more local updates, for the representation extractor (usually corresponds to initial layers) leads to the creation of more generalizable models, particularly for non-iid scenarios. We design a novel Federated Learning with Adaptive Local Steps (FedALS) algorithm based on our generalization bound and representation learning analysis. FedALS employs varying aggregation frequencies for different parts of the model, so reduces the communication cost. The paper is followed with experimental results showing the effectiveness of FedALS.
Authors: Xikun Jiang, He Lyu, Chenhao Ying, Yibin Xu, Boris D\"udder, Yuan Luo
Abstract: With the increasingly widespread application of machine learning, how to strike a balance between protecting the privacy of data and algorithm parameters and ensuring the verifiability of machine learning has always been a challenge. This study explores the intersection of reinforcement learning and data privacy, specifically addressing the Multi-Armed Bandit (MAB) problem with the Upper Confidence Bound (UCB) algorithm. We introduce zkUCB, an innovative algorithm that employs the Zero-Knowledge Succinct Non-Interactive Argument of Knowledge (zk-SNARKs) to enhance UCB. zkUCB is carefully designed to safeguard the confidentiality of training data and algorithmic parameters, ensuring transparent UCB decision-making. Experiments highlight zkUCB's superior performance, attributing its enhanced reward to judicious quantization bit usage that reduces information entropy in the decision-making process. zkUCB's proof size and verification time scale linearly with the execution steps of zkUCB. This showcases zkUCB's adept balance between data security and operational efficiency. This approach contributes significantly to the ongoing discourse on reinforcing data privacy in complex decision-making processes, offering a promising solution for privacy-sensitive applications.
Authors: Fin Amin, Jung-Eun Kim
Abstract: When neural networks are confronted with unfamiliar data that deviate from their training set, this signifies a domain shift. While these networks output predictions on their inputs, they typically fail to account for their level of familiarity with these novel observations. This challenge becomes even more pronounced in resource-constrained settings, such as embedded systems or edge devices. To address such challenges, we aim to recalibrate a neural network's decision boundaries in relation to its cognizance of the data it observes, introducing an approach we coin as certainty distillation. While prevailing works navigate unsupervised domain adaptation (UDA) with the goal of curtailing model entropy, they unintentionally birth models that grapple with calibration inaccuracies - a dilemma we term the over-certainty phenomenon. In this paper, we probe the drawbacks of this traditional learning model. As a solution to the issue, we propose a UDA algorithm that not only augments accuracy but also assures model calibration, all while maintaining suitability for environments with limited computational resources.
Authors: Arsalan Sharifnassab, Sina Ghiassian, Saber Salehkaleybar, Surya Kanoria, Dale Schuurmans
Abstract: We propose Soft Preference Optimization (SPO), a method for aligning generative models, such as Large Language Models (LLMs), with human preferences, without the need for a reward model. SPO optimizes model outputs directly over a preference dataset through a natural loss function that integrates preference loss with a regularization term across the model's entire output distribution rather than limiting it to the preference dataset. Although SPO does not require the assumption of an existing underlying reward model, we demonstrate that, under the Bradley-Terry (BT) model assumption, it converges to a softmax of scaled rewards, with the distribution's "softness" adjustable via the softmax exponent, an algorithm parameter. We showcase SPO's methodology, its theoretical foundation, and its comparative advantages in simplicity, computational efficiency, and alignment precision.
Authors: Zelei Cheng, Xian Wu, Jiahao Yu, Sabrina Yang, Gang Wang, Xinyu Xing
Abstract: Deep reinforcement learning (DRL) is playing an increasingly important role in real-world applications. However, obtaining an optimally performing DRL agent for complex tasks, especially with sparse rewards, remains a significant challenge. The training of a DRL agent can be often trapped in a bottleneck without further progress. In this paper, we propose RICE, an innovative refining scheme for reinforcement learning that incorporates explanation methods to break through the training bottlenecks. The high-level idea of RICE is to construct a new initial state distribution that combines both the default initial states and critical states identified through explanation methods, thereby encouraging the agent to explore from the mixed initial states. Through careful design, we can theoretically guarantee that our refining scheme has a tighter sub-optimality bound. We evaluate RICE in various popular RL environments and real-world applications. The results demonstrate that RICE significantly outperforms existing refining schemes in enhancing agent performance.
Authors: Hangyu Lin, Chen Liu, Chengming Xu, Zhengqi Gao, Yanwei Fu, Yuan Yao
Abstract: Cross-modality distillation arises as an important topic for data modalities containing limited knowledge such as depth maps and high-quality sketches. Such techniques are of great importance, especially for memory and privacy-restricted scenarios where labeled training data is generally unavailable. To solve the problem, existing label-free methods leverage a few pairwise unlabeled data to distill the knowledge by aligning features or statistics between the source and target modalities. For instance, one typically aims to minimize the L2 distance or contrastive loss between the learned features of pairs of samples in the source (e.g. image) and the target (e.g. sketch) modalities. However, most algorithms in this domain only focus on the experimental results but lack theoretical insight. To bridge the gap between the theory and practical method of cross-modality distillation, we first formulate a general framework of cross-modality contrastive distillation (CMCD), built upon contrastive learning that leverages both positive and negative correspondence, towards a better distillation of generalizable features. Furthermore, we establish a thorough convergence analysis that reveals that the distance between source and target modalities significantly impacts the test error on downstream tasks within the target modality which is also validated by the empirical results. Extensive experimental results show that our algorithm outperforms existing algorithms consistently by a margin of 2-3\% across diverse modalities and tasks, covering modalities of image, sketch, depth map, and audio and tasks of recognition and segmentation.
Authors: Wei Ju, Zhengyang Mao, Siyu Yi, Yifang Qin, Yiyang Gu, Zhiping Xiao, Yifan Wang, Xiao Luo, Ming Zhang
Abstract: In this paper, we study semi-supervised graph classification, which aims at accurately predicting the categories of graphs in scenarios with limited labeled graphs and abundant unlabeled graphs. Despite the promising capability of graph neural networks (GNNs), they typically require a large number of costly labeled graphs, while a wealth of unlabeled graphs fail to be effectively utilized. Moreover, GNNs are inherently limited to encoding local neighborhood information using message-passing mechanisms, thus lacking the ability to model higher-order dependencies among nodes. To tackle these challenges, we propose a Hypergraph-Enhanced DuAL framework named HEAL for semi-supervised graph classification, which captures graph semantics from the perspective of the hypergraph and the line graph, respectively. Specifically, to better explore the higher-order relationships among nodes, we design a hypergraph structure learning to adaptively learn complex node dependencies beyond pairwise relations. Meanwhile, based on the learned hypergraph, we introduce a line graph to capture the interaction between hyperedges, thereby better mining the underlying semantic structures. Finally, we develop a relational consistency learning to facilitate knowledge transfer between the two branches and provide better mutual guidance. Extensive experiments on real-world graph datasets verify the effectiveness of the proposed method against existing state-of-the-art methods.
Authors: Junzhuo Chen, Zhixin Lu, Shitong Kang
Abstract: In the realm of machine learning, the KNN classification algorithm is widely recognized for its simplicity and efficiency. However, its sensitivity to the K value poses challenges, especially with small sample sizes or outliers, impacting classification performance. This article introduces a novel KNN-based classifier called LMPHNN (Novel Pseudo Nearest Neighbor Classification Method Using Local Harmonic Mean Distance). LMPHNN leverages harmonic mean distance (HMD) to improve classification performance based on LMPNN rules and HMD. The classifier begins by identifying k nearest neighbors for each class and generates distinct local vectors as prototypes. Pseudo nearest neighbors (PNNs) are then created based on the local mean for each class, determined by comparing the HMD of the sample with the initial k group. Classification is determined by calculating the Euclidean distance between the query sample and PNNs, based on the local mean of these categories. Extensive experiments on various real UCI datasets and combined datasets compare LMPHNN with seven KNN-based classifiers, using precision, recall, accuracy, and F1 as evaluation metrics. LMPHNN achieves an average precision of 97%, surpassing other methods by 14%. The average recall improves by 12%, with an average accuracy enhancement of 5%. Additionally, LMPHNN demonstrates a 13% higher average F1 value compared to other methods. In summary, LMPHNN outperforms other classifiers, showcasing lower sensitivity with small sample sizes.
Authors: Jie Xu, Karthikeyan Saravanan, Rogier van Dalen, Haaris Mehmood, David Tuckey, Mete Ozay
Abstract: Federated learning (FL) allows clients in an Internet of Things (IoT) system to collaboratively train a global model without sharing their local data with a server. However, clients' contributions to the server can still leak sensitive information. Differential privacy (DP) addresses such leakage by providing formal privacy guarantees, with mechanisms that add randomness to the clients' contributions. The randomness makes it infeasible to train large transformer-based models, common in modern IoT systems. In this work, we empirically evaluate the practicality of fine-tuning large scale on-device transformer-based models with differential privacy in a federated learning system. We conduct comprehensive experiments on various system properties for tasks spanning a multitude of domains: speech recognition, computer vision (CV) and natural language understanding (NLU). Our results show that full fine-tuning under differentially private federated learning (DP-FL) generally leads to huge performance degradation which can be alleviated by reducing the dimensionality of contributions through parameter-efficient fine-tuning (PEFT). Our benchmarks of existing DP-PEFT methods show that DP-Low-Rank Adaptation (DP-LoRA) consistently outperforms other methods. An even more promising approach, DyLoRA, which makes the low rank variable, when naively combined with FL would straightforwardly break differential privacy. We therefore propose an adaptation method that can be combined with differential privacy and call it DP-DyLoRA. Finally, we are able to reduce the accuracy degradation and word error rate (WER) increase due to DP to less than 2% and 7% respectively with 1 million clients and a stringent privacy budget of {\epsilon}=2.
Authors: Dongseong Hwang
Abstract: This paper establishes a mathematical foundation for the Adam optimizer, elucidating its connection to natural gradient descent through Riemannian and information geometry. We rigorously analyze the diagonal empirical Fisher information matrix (FIM) in Adam, clarifying all detailed approximations and advocating for the use of log probability functions as loss, which should be based on discrete distributions, due to the limitations of empirical FIM. Our analysis uncovers flaws in the original Adam algorithm, leading to proposed corrections such as enhanced momentum calculations, adjusted bias corrections, adaptive epsilon, and gradient clipping. We refine the weight decay term based on our theoretical framework. Our modified algorithm, Fisher Adam (FAdam), demonstrates superior performance across diverse domains including LLM, ASR, and VQ-VAE, achieving state-of-the-art results in ASR.
Authors: Xiaohua Pan, Weifeng Wu, Peiran Liu, Zhen Li, Peng Lu, Peijian Cao, Jianfeng Zhang, Xianfei Qiu, YangYang Wu
Abstract: In many practical real-world applications, data missing is a very common phenomenon, making the development of data-driven artificial intelligence theory and technology increasingly difficult. Data completion is an important method for missing data preprocessing. Most existing miss-ing data completion models directly use the known information in the missing data set but ignore the impact of the data label information contained in the data set on the missing data completion model. To this end, this paper proposes a missing data completion model SEGAN based on semi-supervised learning, which mainly includes three important modules: generator, discriminator and classifier. In the SEGAN model, the classifier enables the generator to make more full use of known data and its label information when predicting missing data values. In addition, the SE-GAN model introduces a missing hint matrix to allow the discriminator to more effectively distinguish between known data and data filled by the generator. This paper theoretically proves that the SEGAN model that introduces a classifier and a missing hint matrix can learn the real known data distribution characteristics when reaching Nash equilibrium. Finally, a large number of experiments were conducted in this article, and the experimental results show that com-pared with the current state-of-the-art multivariate data completion method, the performance of the SEGAN model is improved by more than 3%.
Authors: Peiwang Tang, Weitai Zhang
Abstract: Recent studies have attempted to refine the Transformer architecture to demonstrate its effectiveness in Long-Term Time Series Forecasting (LTSF) tasks. Despite surpassing many linear forecasting models with ever-improving performance, we remain skeptical of Transformers as a solution for LTSF. We attribute the effectiveness of these models largely to the adopted Patch mechanism, which enhances sequence locality to an extent yet fails to fully address the loss of temporal information inherent to the permutation-invariant self-attention mechanism. Further investigation suggests that simple linear layers augmented with the Patch mechanism may outperform complex Transformer-based LTSF models. Moreover, diverging from models that use channel independence, our research underscores the importance of cross-variable interactions in enhancing the performance of multivariate time series forecasting. The interaction information between variables is highly valuable but has been misapplied in past studies, leading to suboptimal cross-variable models. Based on these insights, we propose a novel and simple Patch-based Decomposed MLP (PDMLP) for LTSF tasks. Specifically, we employ simple moving averages to extract smooth components and noise-containing residuals from time series data, engaging in semantic information interchange through channel mixing and specializing in random noise with channel independence processing. The PDMLP model consistently achieves state-of-the-art results on several real-world datasets. We hope this surprising finding will spur new research directions in the LTSF field and pave the way for more efficient and concise solutions.
Authors: Jiali Cui, Tian Han
Abstract: This work studies the learning problem of the energy-based prior model and the multi-layer generator model. The multi-layer generator model, which contains multiple layers of latent variables organized in a top-down hierarchical structure, typically assumes the Gaussian prior model. Such a prior model can be limited in modelling expressivity, which results in a gap between the generator posterior and the prior model, known as the prior hole problem. Recent works have explored learning the energy-based (EBM) prior model as a second-stage, complementary model to bridge the gap. However, the EBM defined on a multi-layer latent space can be highly multi-modal, which makes sampling from such marginal EBM prior challenging in practice, resulting in ineffectively learned EBM. To tackle the challenge, we propose to leverage the diffusion probabilistic scheme to mitigate the burden of EBM sampling and thus facilitate EBM learning. Our extensive experiments demonstrate a superior performance of our diffusion-learned EBM prior on various challenging tasks.
Authors: Xingtong Yu, Chang Zhou, Yuan Fang, Xinming Zhang
Abstract: Given the ubiquity of graph data, it is intriguing to ask: Is it possible to train a graph foundation model on a broad range of graph data across diverse domains? A major hurdle toward this goal lies in the fact that graphs from different domains often exhibit profoundly divergent characteristics. Although there have been some initial efforts in integrating multi-domain graphs for pre-training, they primarily rely on textual descriptions to align the graphs, limiting their application to text-attributed graphs. Moreover, different source domains may conflict or interfere with each other, and their relevance to the target domain can vary significantly. To address these issues, we propose MDGPT, a text free Multi-Domain Graph Pre-Training and adaptation framework designed to exploit multi-domain knowledge for graph learning. First, we propose a set of domain tokens to to align features across source domains for synergistic pre-training. Second, we propose a dual prompts, consisting of a unifying prompt and a mixing prompt, to further adapt the target domain with unified multi-domain knowledge and a tailored mixture of domain-specific knowledge. Finally, we conduct extensive experiments involving six public datasets to evaluate and analyze MDGPT, which outperforms prior art by up to 37.9%.
Authors: Xingtong Yu, Zhenghao Liu, Yuan Fang, Xinming Zhang
Abstract: Dynamic graphs are pervasive in the real world, modeling dynamic relations between objects across various fields. For dynamic graph modeling, dynamic graph neural networks (DGNNs) have emerged as a mainstream technique, which are generally pre-trained on the link prediction task, leaving a significant gap from the objectives of downstream tasks such as node classification. To bridge the gap, prompt-based learning has gained traction on graphs. However, existing efforts focus on static graphs, neglecting the evolution of dynamic graphs. In this paper, we propose DyGPrompt, a novel pre-training and prompting framework for dynamic graph modeling. First, we design dual prompts to address the gap in both task objectives and dynamic variations across pre-training and downstream tasks. Second, we recognize that node and time features mutually characterize each other, and propose dual condition-nets to model the evolving node-time patterns in downstream tasks. Finally, we thoroughly evaluate and analyze DyGPrompt through extensive experiments on three public datasets.
Authors: Leo Feng, Frederick Tung, Hossein Hajimirsadeghi, Mohamed Osama Ahmed, Yoshua Bengio, Greg Mori
Abstract: The advent of Transformers marked a significant breakthrough in sequence modelling, providing a highly performant architecture capable of leveraging GPU parallelism. However, Transformers are computationally expensive at inference time, limiting their applications, particularly in low-resource settings (e.g., mobile and embedded devices). Addressing this, we (1) begin by showing that attention can be viewed as a special Recurrent Neural Network (RNN) with the ability to compute its \textit{many-to-one} RNN output efficiently. We then (2) show that popular attention-based models such as Transformers can be viewed as RNN variants. However, unlike traditional RNNs (e.g., LSTMs), these models cannot be updated efficiently with new tokens, an important property in sequence modelling. Tackling this, we (3) introduce a new efficient method of computing attention's \textit{many-to-many} RNN output based on the parallel prefix scan algorithm. Building on the new attention formulation, we (4) introduce \textbf{Aaren}, an attention-based module that can not only (i) be trained in parallel (like Transformers) but also (ii) be updated efficiently with new tokens, requiring only constant memory for inferences (like traditional RNNs). Empirically, we show Aarens achieve comparable performance to Transformers on $38$ datasets spread across four popular sequential problem settings: reinforcement learning, event forecasting, time series classification, and time series forecasting tasks while being more time and memory-efficient.
Authors: Johannes Ackermann, Takayuki Osa, Masashi Sugiyama
Abstract: Current Reinforcement Learning (RL) is often limited by the large amount of data needed to learn a successful policy. Offline RL aims to solve this issue by using transitions collected by a different behavior policy. We address a novel Offline RL problem setting in which, while collecting the dataset, the transition and reward functions gradually change between episodes but stay constant within each episode. We propose a method based on Contrastive Predictive Coding that identifies this non-stationarity in the offline dataset, accounts for it when training a policy, and predicts it during evaluation. We analyze our proposed method and show that it performs well in simple continuous control tasks and challenging, high-dimensional locomotion tasks. We show that our method often achieves the oracle performance and performs better than baselines.
Authors: Qingyuan Li, Ran Meng, Yiduo Li, Bo Zhang, Yifan Lu, Yerui Sun, Lin Ma, Yuchen Xie
Abstract: We introduce Integer Scale, a novel post-training quantization scheme for large language models that effectively resolves the inference bottleneck in current fine-grained quantization approaches while maintaining similar accuracies. Integer Scale is a free lunch as it requires no extra calibration or fine-tuning which will otherwise incur additional costs. It can be used plug-and-play for most fine-grained quantization methods. Its integration results in at most 1.85x end-to-end speed boost over the original counterpart with comparable accuracy. Additionally, due to the orchestration of the proposed Integer Scale and fine-grained quantization, we resolved the quantization difficulty for Mixtral-8x7B and LLaMA-3 models with negligible performance degradation, and it comes with an end-to-end speed boost of 2.13x, and 2.31x compared with their FP16 versions respectively.
Authors: Ziqi Shi, Fan Lyu, Ye Liu, Fanhua Shang, Fuyuan Hu, Wei Feng, Zhang Zhang, Liang Wang
Abstract: Continual Test-Time Adaptation (CTTA) is an emerging and challenging task where a model trained in a source domain must adapt to continuously changing conditions during testing, without access to the original source data. CTTA is prone to error accumulation due to uncontrollable domain shifts, leading to blurred decision boundaries between categories. Existing CTTA methods primarily focus on suppressing domain shifts, which proves inadequate during the unsupervised test phase. In contrast, we introduce a novel approach that guides rather than suppresses these shifts. Specifically, we propose $\textbf{C}$ontrollable $\textbf{Co}$ntinual $\textbf{T}$est-$\textbf{T}$ime $\textbf{A}$daptation (C-CoTTA), which explicitly prevents any single category from encroaching on others, thereby mitigating the mutual influence between categories caused by uncontrollable shifts. Moreover, our method reduces the sensitivity of model to domain transformations, thereby minimizing the magnitude of category shifts. Extensive quantitative experiments demonstrate the effectiveness of our method, while qualitative analyses, such as t-SNE plots, confirm the theoretical validity of our approach.
Authors: Yu Zhe, Rei Nagaike, Daiki Nishiyama, Kazuto Fukuchi, Jun Sakuma
Abstract: Deep learning models are susceptible to adversarial attacks, where slight perturbations to input data lead to misclassification. Adversarial attacks become increasingly effective with access to information about the targeted classifier. In the context of multi-task learning, where a single model learns multiple tasks simultaneously, attackers may aim to exploit vulnerabilities in specific tasks with limited information. This paper investigates the feasibility of attacking hidden tasks within multi-task classifiers, where model access regarding the hidden target task and labeled data for the hidden target task are not available, but model access regarding the non-target tasks is available. We propose a novel adversarial attack method that leverages knowledge from non-target tasks and the shared backbone network of the multi-task model to force the model to forget knowledge related to the target task. Experimental results on CelebA and DeepFashion datasets demonstrate the effectiveness of our method in degrading the accuracy of hidden tasks while preserving the performance of visible tasks, contributing to the understanding of adversarial vulnerabilities in multi-task classifiers.
Authors: Jiantong Jiang, Zeyi Wen, Peiyu Yang, Atif Mansoor, Ajmal Mian
Abstract: Probabilistic graphical models (PGMs) serve as a powerful framework for modeling complex systems with uncertainty and extracting valuable insights from data. However, users face challenges when applying PGMs to their problems in terms of efficiency and usability. This paper presents Fast-PGM, an efficient and open-source library for PGM learning and inference. Fast-PGM supports comprehensive tasks on PGMs, including structure and parameter learning, as well as exact and approximate inference, and enhances efficiency of the tasks through computational and memory optimizations and parallelization techniques. Concurrently, Fast-PGM furnishes developers with flexible building blocks, furnishes learners with detailed documentation, and affords non-experts user-friendly interfaces, thereby ameliorating the usability of PGMs to users across a spectrum of expertise levels. The source code of Fast-PGM is available at https://github.com/jjiantong/FastPGM.
Authors: Michael Munn, Benoit Dherin, Javier Gonzalvo
Abstract: Many of the recent remarkable advances in computer vision and language models can be attributed to the success of transfer learning via the pre-training of large foundation models. However, a theoretical framework which explains this empirical success is incomplete and remains an active area of research. Flatness of the loss surface and neural collapse have recently emerged as useful pre-training metrics which shed light on the implicit biases underlying pre-training. In this paper, we explore the geometric complexity of a model's learned representations as a fundamental mechanism that relates these two concepts. We show through experiments and theory that mechanisms which affect the geometric complexity of the pre-trained network also influence the neural collapse. Furthermore, we show how this effect of the geometric complexity generalizes to the neural collapse of new classes as well, thus encouraging better performance on downstream tasks, particularly in the few-shot setting.
Authors: Roel Bouman, Linda Schmeitz, Luco Buise, Jacco Heres, Yuliya Shapovalova, Tom Heskes
Abstract: In this paper we present novel methodology for automatic anomaly and switch event filtering to improve load estimation in power grid systems. By leveraging unsupervised methods with supervised optimization, our approach prioritizes interpretability while ensuring robust and generalizable performance on unseen data. Through experimentation, a combination of binary segmentation for change point detection and statistical process control for anomaly detection emerges as the most effective strategy, specifically when ensembled in a novel sequential manner. Results indicate the clear wasted potential when filtering is not applied. The automatic load estimation is also fairly accurate, with approximately 90% of estimates falling within a 10% error margin, with only a single significant failure in both the minimum and maximum load estimates across 60 measurements in the test set. Our methodology's interpretability makes it particularly suitable for critical infrastructure planning, thereby enhancing decision-making processes.
Authors: Gabriele Dominici, Pietro Barbiero, Mateo Espinosa Zarlenga, Alberto Termine, Martin Gjoreski, Giuseppe Marra, Marc Langheinrich
Abstract: Causal opacity denotes the difficulty in understanding the "hidden" causal structure underlying a deep neural network's (DNN) reasoning. This leads to the inability to rely on and verify state-of-the-art DNN-based systems especially in high-stakes scenarios. For this reason, causal opacity represents a key open challenge at the intersection of deep learning, interpretability, and causality. This work addresses this gap by introducing Causal Concept Embedding Models (Causal CEMs), a class of interpretable models whose decision-making process is causally transparent by design. The results of our experiments show that Causal CEMs can: (i) match the generalization performance of causally-opaque models, (ii) support the analysis of interventional and counterfactual scenarios, thereby improving the model's causal interpretability and supporting the effective verification of its reliability and fairness, and (iii) enable human-in-the-loop corrections to mispredicted intermediate reasoning steps, boosting not just downstream accuracy after corrections but also accuracy of the explanation provided for a specific instance.
Authors: Mohammed Nowaz Rabbani Chowdhury, Meng Wang, Kaoutar El Maghraoui, Naigang Wang, Pin-Yu Chen, Christopher Carothers
Abstract: The sparsely gated mixture of experts (MoE) architecture sends different inputs to different subnetworks, i.e., experts, through trainable routers. MoE reduces the training computation significantly for large models, but its deployment can be still memory or computation expensive for some downstream tasks. Model pruning is a popular approach to reduce inference computation, but its application in MoE architecture is largely unexplored. To the best of our knowledge, this paper provides the first provably efficient technique for pruning experts in finetuned MoE models. We theoretically prove that prioritizing the pruning of the experts with a smaller change of the routers l2 norm from the pretrained model guarantees the preservation of test accuracy, while significantly reducing the model size and the computational requirements. Although our theoretical analysis is centered on binary classification tasks on simplified MoE architecture, our expert pruning method is verified on large vision MoE models such as VMoE and E3MoE finetuned on benchmark datasets such as CIFAR10, CIFAR100, and ImageNet.
Authors: Yidong Ouyang, Liyan Xie, Hongyuan Zha, Guang Cheng
Abstract: Diffusion models, a specific type of generative model, have achieved unprecedented performance in recent years and consistently produce high-quality synthetic samples. A critical prerequisite for their notable success lies in the presence of a substantial number of training samples, which can be impractical in real-world applications due to high collection costs or associated risks. Consequently, various finetuning and regularization approaches have been proposed to transfer knowledge from existing pre-trained models to specific target domains with limited data. This paper introduces the Transfer Guided Diffusion Process (TGDP), a novel approach distinct from conventional finetuning and regularization methods. We prove that the optimal diffusion model for the target domain integrates pre-trained diffusion models on the source domain with additional guidance from a domain classifier. We further extend TGDP to a conditional version for modeling the joint distribution of data and its corresponding labels, together with two additional regularization terms to enhance the model performance. We validate the effectiveness of TGDP on Gaussian mixture simulations and on real electrocardiogram (ECG) datasets.
Authors: Filip Postepski, Grzegorz M. Wojcik, Krzysztof Wrobel, Andrzej Kawiak, Katarzyna Zemla, Grzegorz Sedek
Abstract: The Guided Imagery technique is reported to be used by therapists all over the world in order to increase the comfort of patients suffering from a variety of disorders from mental to oncology ones and proved to be successful in numerous of ways. Possible support for the therapists can be estimation of the time at which subject goes into deep relaxation. This paper presents the results of the investigations of a cohort of 26 students exposed to Guided Imagery relaxation technique and mental task workloads conducted with the use of dense array electroencephalographic amplifier. The research reported herein aimed at verification whether it is possible to detect differences between those two states and to classify them using deep learning methods and recurrent neural networks such as EEGNet, Long Short-Term Memory-based classifier, 1D Convolutional Neural Network and hybrid model of 1D Convolutional Neural Network and Long Short-Term Memory. The data processing pipeline was presented from the data acquisition, through the initial data cleaning, preprocessing and postprocessing. The classification was based on two datasets: one of them using 26 so-called cognitive electrodes and the other one using signal collected from 256 channels. So far there have not been such comparisons in the application being discussed. The classification results are presented by the validation metrics such as: accuracy, recall, precision, F1-score and loss for each case. It turned out that it is not necessary to collect signals from all electrodes as classification of the cognitive ones gives the results similar to those obtained for the full signal and extending input to 256 channels does not add much value. In Disscussion there were proposed an optimal classifier as well as some suggestions concerning the prospective development of the project.
Authors: Maximilian Granz, Manuel Heurich, Tim Landgraf
Abstract: Recent advances in out-of-distribution (OOD) detection on image data show that pre-trained neural network classifiers can separate in-distribution (ID) from OOD data well, leveraging the class-discriminative ability of the model itself. Methods have been proposed that either use logit information directly or that process the model's penultimate layer activations. With "WeiPer", we introduce perturbations of the class projections in the final fully connected layer which creates a richer representation of the input. We show that this simple trick can improve the OOD detection performance of a variety of methods and additionally propose a distance-based method that leverages the properties of the augmented WeiPer space. We achieve state-of-the-art OOD detection results across multiple benchmarks of the OpenOOD framework, especially pronounced in difficult settings in which OOD samples are positioned close to the training set distribution. We support our findings with theoretical motivations and empirical observations, and run extensive ablations to provide insights into why WeiPer works.
Authors: Cedric Kulbach, Lucas Cazzonelli, Hoang-Anh Ngo, Minh-Huong Le-Nguyen, Albert Bifet
Abstract: Machine learning algorithms have become indispensable in today's world. They support and accelerate the way we make decisions based on the data at hand. This acceleration means that data structures that were valid at one moment could no longer be valid in the future. With these changing data structures, it is necessary to adapt machine learning (ML) systems incrementally to the new data. This is done with the use of online learning or continuous ML technologies. While deep learning technologies have shown exceptional performance on predefined datasets, they have not been widely applied to online, streaming, and continuous learning. In this retrospective of our tutorial titled Opportunities and Challenges of Online Deep Learning held at ECML PKDD 2023, we provide a brief overview of the opportunities but also the potential pitfalls for the application of neural networks in online learning environments using the frameworks River and Deep-River.
Authors: Anand Gopalakrishnan, Aleksandar Stani\'c, J\"urgen Schmidhuber, Michael Curtis Mozer
Abstract: Current state-of-the-art synchrony-based models encode object bindings with complex-valued activations and compute with real-valued weights in feedforward architectures. We argue for the computational advantages of a recurrent architecture with complex-valued weights. We propose a fully convolutional autoencoder, SynCx, that performs iterative constraint satisfaction: at each iteration, a hidden layer bottleneck encodes statistically regular configurations of features in particular phase relationships; over iterations, local constraints propagate and the model converges to a globally consistent configuration of phase assignments. Binding is achieved simply by the matrix-vector product operation between complex-valued weights and activations, without the need for additional mechanisms that have been incorporated into current synchrony-based models. SynCx outperforms or is strongly competitive with current models for unsupervised object discovery. SynCx also avoids certain systematic grouping errors of current models, such as the inability to separate similarly colored objects without additional supervision.
Authors: ShengYun Peng, Pin-Yu Chen, Matthew Hull, Duen Horng Chau
Abstract: Safety alignment is the key to guiding the behaviors of large language models (LLMs) that are in line with human preferences and restrict harmful behaviors at inference time, but recent studies show that it can be easily compromised by finetuning with only a few adversarially designed training examples. We aim to measure the risks in finetuning LLMs through navigating the LLM safety landscape. We discover a new phenomenon observed universally in the model parameter space of popular open-source LLMs, termed as "safety basin": randomly perturbing model weights maintains the safety level of the original aligned model in its local neighborhood. Our discovery inspires us to propose the new VISAGE safety metric that measures the safety in LLM finetuning by probing its safety landscape. Visualizing the safety landscape of the aligned model enables us to understand how finetuning compromises safety by dragging the model away from the safety basin. LLM safety landscape also highlights the system prompt's critical role in protecting a model, and that such protection transfers to its perturbed variants within the safety basin. These observations from our safety landscape research provide new insights for future work on LLM safety community.
Authors: Cencheng Shen, Jaewon Chung, Ronak Mehta, Ting Xu, Joshua T. Vogelstein
Abstract: Temporal data are increasingly prevalent in modern data science. A fundamental question is whether two time series are related or not. Existing approaches often have limitations, such as relying on parametric assumptions, detecting only linear associations, and requiring multiple tests and corrections. While many non-parametric and universally consistent dependence measures have recently been proposed, directly applying them to temporal data can inflate the p-value and result in an invalid test. To address these challenges, this paper introduces the temporal dependence statistic with block permutation to test independence between temporal data. Under proper assumptions, the proposed procedure is asymptotically valid and universally consistent for testing independence between stationary time series, and capable of estimating the optimal dependence lag that maximizes the dependence. Moreover, it is compatible with a rich family of distance and kernel based dependence measures, eliminates the need for multiple testing, and exhibits excellent testing power in various simulation settings.
Authors: Guanlin Li, Guowen Xu, Han Qiu, Shangwei Guo, Run Wang, Jiwei Li, Tianwei Zhang, Rongxing Lu
Abstract: Generative Adversarial Networks (GANs) have been widely used in various application scenarios. Since the production of a commercial GAN requires substantial computational and human resources, the copyright protection of GANs is urgently needed. This paper presents a novel fingerprinting scheme for the Intellectual Property (IP) protection of image-to-image GANs based on a trusted third party. We break through the stealthiness and robustness bottlenecks suffered by previous fingerprinting methods for classification models being naively transferred to GANs. Specifically, we innovatively construct a composite deep learning model from the target GAN and a classifier. Then we generate fingerprint samples from this composite model, and embed them in the classifier for effective ownership verification. This scheme inspires some concrete methodologies to practically protect the modern image-to-image translation GANs. Theoretical analysis proves that these methods can satisfy different security requirements necessary for IP protection. We also conduct extensive experiments to show that our solutions outperform existing strategies.
Authors: Qinggang Sun, Kejun Wang
Abstract: The separation of single-channel underwater acoustic signals is a challenging problem with practical significance. Few existing studies focus on the source separation problem with unknown numbers of signals, and how to evaluate the performance of the systems is not yet clear. In this paper, a deep learning-based simultaneous separating solution with a fixed number of output channels equal to the maximum number of possible targets is proposed to address these two problems. This solution avoids the dimensional disaster caused by the permutation problem induced by the alignment of outputs to targets. Specifically, we propose a two-step learning-based separation model with a separator-decoder structure. A performance evaluation method with two quantitative metrics of the separation system for situations with mute channels in the output channels that do not contain target signals is also proposed. Experiments conducted on simulated mixtures of radiated ship noise show that the proposed solution can achieve similar separation performance to that attained with a known number of signals. The proposed separation model with separator-decoder structure achieved competitive performance as two models developed for known numbers of signals, which is highly explainable and extensible and gets the state of the art under this framework.
Authors: Zaid Harchaoui, Sewoong Oh, Soumik Pal, Raghav Somani, Raghavendra Tripathi
Abstract: We consider stochastic gradient descents on the space of large symmetric matrices of suitable functions that are invariant under permuting the rows and columns using the same permutation. We establish deterministic limits of these random curves as the dimensions of the matrices go to infinity while the entries remain bounded. Under a ``small noise'' assumption the limit is shown to be the gradient flow of functions on graphons whose existence was established in~\cite{oh2021gradient}. We also consider limits of stochastic gradient descents with added properly scaled reflected Brownian noise. The limiting curve of graphons is characterized by a family of stochastic differential equations with reflections and can be thought of as an extension of the classical McKean-Vlasov limit for interacting diffusions to the graphon setting. The proofs introduce a family of infinite-dimensional exchangeable arrays of reflected diffusions and a novel notion of propagation of chaos for large matrices of diffusions converging to such arrays in a suitable sense.
Authors: Abdulla Jasem Almansoori, Samuel Horv\'ath, Martin Tak\'a\v{c}
Abstract: Federated learning has become a popular machine learning paradigm with many potential real-life applications, including recommendation systems, the Internet of Things (IoT), healthcare, and self-driving cars. Though most current applications focus on classification-based tasks, learning personalized generative models remains largely unexplored, and their benefits in the heterogeneous setting still need to be better understood. This work proposes a novel architecture combining global client-agnostic and local client-specific generative models. We show that using standard techniques for training federated models, our proposed model achieves privacy and personalization by implicitly disentangling the globally consistent representation (i.e. content) from the client-dependent variations (i.e. style). Using such decomposition, personalized models can generate locally unseen labels while preserving the given style of the client and can predict the labels for all clients with high accuracy by training a simple linear classifier on the global content features. Furthermore, disentanglement enables other essential applications, such as data anonymization, by sharing only the content. Extensive experimental evaluation corroborates our findings, and we also discuss a theoretical motivation for the proposed approach.
Authors: Hans W. A. Hanley, Zakir Durumeric
Abstract: In response to disinformation and propaganda from Russian online media following the invasion of Ukraine, Russian media outlets such as Russia Today and Sputnik News were banned throughout Europe. To maintain viewership, many of these Russian outlets began to heavily promote their content on messaging services like Telegram. In this work, we study how 16 Russian media outlets interacted with and utilized 732 Telegram channels throughout 2022. Leveraging the foundational model MPNet, DP-means clustering, and Hawkes processes, we trace how narratives spread between news sites and Telegram channels. We show that news outlets not only propagate existing narratives through Telegram but that they source material from the messaging platform. For example, across the websites in our study, between 2.3% (ura.news) and 26.7% (ukraina.ru) of articles discussed content that originated/resulted from activity on Telegram. Finally, tracking the spread of individual topics, we measure the rate at which news outlets and Telegram channels disseminate content within the Russian media ecosystem, finding that websites like ura.news and Telegram channels such as @genshab are the most effective at disseminating their content.
Authors: Oren Yuval, Saharon Rosset
Abstract: We present a methodology for using unlabeled data to design semi supervised learning (SSL) methods that improve the prediction performance of supervised learning for regression tasks. The main idea is to design different mechanisms for integrating the unlabeled data, and include in each of them a mixing parameter $\alpha$, controlling the weight given to the unlabeled data. Focusing on Generalized Linear Models (GLM) and linear interpolators classes of models, we analyze the characteristics of different mixing mechanisms, and prove that in all cases, it is invariably beneficial to integrate the unlabeled data with some nonzero mixing ratio $\alpha>0$, in terms of predictive performance. Moreover, we provide a rigorous framework to estimate the best mixing ratio $\alpha^*$ where mixed SSL delivers the best predictive performance, while using the labeled and unlabeled data on hand. The effectiveness of our methodology in delivering substantial improvement compared to the standard supervised models, in a variety of settings, is demonstrated empirically through extensive simulation, in a manner that supports the theoretical analysis. We also demonstrate the applicability of our methodology (with some intuitive modifications) to improve more complex models, such as deep neural networks, in real-world regression tasks.
Authors: Michael Hersche, Aleksandar Terzic, Geethan Karunaratne, Jovin Langenegger, Ang\'eline Pouget, Giovanni Cherubini, Luca Benini, Abu Sebastian, Abbas Rahimi
Abstract: Distributed sparse block codes (SBCs) exhibit compact representations for encoding and manipulating symbolic data structures using fixed-width vectors. One major challenge however is to disentangle, or factorize, the distributed representation of data structures into their constituent elements without having to search through all possible combinations. This factorization becomes more challenging when SBCs vectors are noisy due to perceptual uncertainty and approximations made by modern neural networks to generate the query SBCs vectors. To address these challenges, we first propose a fast and highly accurate method for factorizing a more flexible and hence generalized form of SBCs, dubbed GSBCs. Our iterative factorizer introduces a threshold-based nonlinear activation, conditional random sampling, and an $\ell_\infty$-based similarity metric. Secondly, the proposed factorizer maintains a high accuracy when queried by noisy product vectors generated using deep convolutional neural networks (CNNs). This facilitates its application in replacing the large fully connected layer (FCL) in CNNs, whereby $C$ trainable class vectors, or attribute combinations, can be implicitly represented by our factorizer having $F$-factor codebooks, each with $\sqrt[\leftroot{-2}\uproot{2}F]{C}$ fixed codevectors. We provide a methodology to flexibly integrate our factorizer in the classification layer of CNNs with a novel loss function. With this integration, the convolutional layers can generate a noisy product vector that our factorizer can still decode, whereby the decoded factors can have different interpretations based on downstream tasks. We demonstrate the feasibility of our method on four deep CNN architectures over CIFAR-100, ImageNet-1K, and RAVEN datasets. In all use cases, the number of parameters and operations are notably reduced compared to the FCL.
Authors: Shady Abu-Hussein, Raja Giryes
Abstract: Denoising Diffusion Probabilistic Models (DDPM) have recently gained significant attention. DDPMs compose a Markovian process that begins in the data domain and gradually adds noise until reaching pure white noise. DDPMs generate high-quality samples from complex data distributions by defining an inverse process and training a deep neural network to learn this mapping. However, these models are inefficient because they require many diffusion steps to produce aesthetically pleasing samples. Additionally, unlike generative adversarial networks (GANs), the latent space of diffusion models is less interpretable. In this work, we propose to generalize the denoising diffusion process into an Upsampling Diffusion Probabilistic Model (UDPM). In the forward process, we reduce the latent variable dimension through downsampling, followed by the traditional noise perturbation. As a result, the reverse process gradually denoises and upsamples the latent variable to produce a sample from the data distribution. We formalize the Markovian diffusion processes of UDPM and demonstrate its generation capabilities on the popular FFHQ, AFHQv2, and CIFAR10 datasets. UDPM generates images with as few as three network evaluations, whose overall computational cost is less than a single DDPM or EDM step, while achieving an FID score of 6.86. This surpasses current state-of-the-art efficient diffusion models that use a single denoising step for sampling. Additionally, UDPM offers an interpretable and interpolable latent space, which gives it an advantage over traditional DDPMs. Our code is available online: \url{https://github.com/shadyabh/UDPM/}
Authors: Lijun Zhang, Haomin Bai, Wei-Wei Tu, Ping Yang, Yao Hu
Abstract: While traditional distributionally robust optimization (DRO) aims to minimize the maximal risk over a set of distributions, Agarwal and Zhang (2022) recently proposed a variant that replaces risk with excess risk. Compared to DRO, the new formulation$\unicode{x2013}$minimax excess risk optimization (MERO) has the advantage of suppressing the effect of heterogeneous noise in different distributions. However, the choice of excess risk leads to a very challenging minimax optimization problem, and currently there exists only an inefficient algorithm for empirical MERO. In this paper, we develop efficient stochastic approximation approaches which directly target MERO. Specifically, we leverage techniques from stochastic convex optimization to estimate the minimal risk of every distribution, and solve MERO as a stochastic convex-concave optimization (SCCO) problem with biased gradients. The presence of bias makes existing theoretical guarantees of SCCO inapplicable, and fortunately, we demonstrate that the bias, caused by the estimation error of the minimal risk, is under-control. Thus, MERO can still be optimized with a nearly optimal convergence rate. Moreover, we investigate a practical scenario where the quantity of samples drawn from each distribution may differ, and propose a stochastic approach that delivers distribution-dependent convergence rates.
Authors: Anish Dhir, Samuel Power, Mark van der Wilk
Abstract: Much of the causal discovery literature prioritises guaranteeing the identifiability of causal direction in statistical models. For structures within a Markov equivalence class, this requires strong assumptions which may not hold in real-world datasets, ultimately limiting the usability of these methods. Building on previous attempts, we show how to incorporate causal assumptions within the Bayesian framework. Identifying causal direction then becomes a Bayesian model selection problem. This enables us to construct models with realistic assumptions, and consequently allows for the differentiation between Markov equivalent causal structures. We analyse why Bayesian model selection works in situations where methods based on maximum likelihood fail. To demonstrate our approach, we construct a Bayesian non-parametric model that can flexibly model the joint distribution. We then outperform previous methods on a wide range of benchmark datasets with varying data generating assumptions.
Authors: Theshani Nuradha, Ziv Goldfeld, Mark M. Wilde
Abstract: We propose a versatile privacy framework for quantum systems, termed quantum pufferfish privacy (QPP). Inspired by classical pufferfish privacy, our formulation generalizes and addresses limitations of quantum differential privacy by offering flexibility in specifying private information, feasible measurements, and domain knowledge. We show that QPP can be equivalently formulated in terms of the Datta-Leditzky information spectrum divergence, thus providing the first operational interpretation thereof. We reformulate this divergence as a semi-definite program and derive several properties of it, which are then used to prove convexity, composability, and post-processing of QPP mechanisms. Parameters that guarantee QPP of the depolarization mechanism are also derived. We analyze the privacy-utility tradeoff of general QPP mechanisms and, again, study the depolarization mechanism as an explicit instance. The QPP framework is then applied to privacy auditing for identifying privacy violations via a hypothesis testing pipeline that leverages quantum algorithms. Connections to quantum fairness and other quantum divergences are also explored and several variants of QPP are examined.
Authors: Wenjun Jiang, Tianlong Fan, Changhao Li, Chuanfu Zhang, Tao Zhang, Zong-fu Luo
Abstract: Connectivity robustness, a crucial aspect for understanding, optimizing, and repairing complex networks, has traditionally been evaluated through time-consuming and often impractical simulations. Fortunately, machine learning provides a new avenue for addressing this challenge. However, several key issues remain unresolved, including the performance in more general edge removal scenarios, capturing robustness through attack curves instead of directly training for robustness, scalability of predictive tasks, and transferability of predictive capabilities. In this paper, we address these challenges by designing a convolutional neural networks (CNN) model with spatial pyramid pooling networks (SPP-net), adapting existing evaluation metrics, redesigning the attack modes, introducing appropriate filtering rules, and incorporating the value of robustness as training data. The results demonstrate the thoroughness of the proposed CNN framework in addressing the challenges of high computational time across various network types, failure component types and failure scenarios. However, the performance of the proposed CNN model varies: for evaluation tasks that are consistent with the trained network type, the proposed CNN model consistently achieves accurate evaluations of both attack curves and robustness values across all removal scenarios. When the predicted network type differs from the trained network, the CNN model still demonstrates favorable performance in the scenario of random node failure, showcasing its scalability and performance transferability. Nevertheless, the performance falls short of expectations in other removal scenarios. This observed scenario-sensitivity in the evaluation of network features has been overlooked in previous studies and necessitates further attention and optimization. Lastly, we discuss important unresolved questions and further investigation.
Authors: Jennifer A Bishop, Qianqian Xie, Sophia Ananiadou
Abstract: Maintaining factual consistency is a critical issue in abstractive text summarisation, however, it cannot be assessed by traditional automatic metrics used for evaluating text summarisation, such as ROUGE scoring. Recent efforts have been devoted to developing improved metrics for measuring factual consistency using pre-trained language models, but these metrics have restrictive token limits, and are therefore not suitable for evaluating long document text summarisation. Moreover, there is limited research and resources available for evaluating whether existing automatic evaluation metrics are fit for purpose when applied in long document settings. In this work, we evaluate the efficacy of automatic metrics for assessing the factual consistency of long document text summarisation. We create a human-annotated data set for evaluating automatic factuality metrics, LongSciVerify, which contains fine-grained factual consistency annotations for long document summaries from the scientific domain. We also propose a new evaluation framework, LongDocFACTScore, which is suitable for evaluating long document summarisation. This framework allows metrics to be efficiently extended to any length document and outperforms existing state-of-the-art metrics in its ability to correlate with human measures of factuality when used to evaluate long document summarisation data sets. We make our code and LongSciVerify data set publicly available: https://github.com/jbshp/LongDocFACTScore.
Authors: Jannis Bulian, Mike S. Sch\"afer, Afra Amini, Heidi Lam, Massimiliano Ciaramita, Ben Gaiarin, Michelle Chen H\"ubscher, Christian Buck, Niels G. Mede, Markus Leippold, Nadine Strau{\ss}
Abstract: As Large Language Models (LLMs) rise in popularity, it is necessary to assess their capability in critically relevant domains. We present a comprehensive evaluation framework, grounded in science communication research, to assess LLM responses to questions about climate change. Our framework emphasizes both presentational and epistemological adequacy, offering a fine-grained analysis of LLM generations spanning 8 dimensions and 30 issues. Our evaluation task is a real-world example of a growing number of challenging problems where AI can complement and lift human performance. We introduce a novel protocol for scalable oversight that relies on AI Assistance and raters with relevant education. We evaluate several recent LLMs on a set of diverse climate questions. Our results point to a significant gap between surface and epistemological qualities of LLMs in the realm of climate communication.
Authors: Jaewoo Lee, Jaehong Yoon, Wonjae Kim, Yunji Kim, Sung Ju Hwang
Abstract: Continuously learning a variety of audio-video semantics over time is crucial for audio-related reasoning tasks in our ever-evolving world. However, this is a nontrivial problem and poses two critical challenges: sparse spatio-temporal correlation between audio-video pairs and multimodal correlation overwriting that forgets audio-video relations. To tackle this problem, we propose a new continual audio-video pre-training method with two novel ideas: (1) Localized Patch Importance Scoring: we introduce a multimodal encoder to determine the importance score for each patch, emphasizing semantically intertwined audio-video patches. (2) Replay-guided Correlation Assessment: to reduce the corruption of previously learned audiovisual knowledge due to drift, we propose to assess the correlation of the current patches on the past steps to identify the patches exhibiting high correlations with the past steps. Based on the results from the two ideas, we perform probabilistic patch selection for effective continual audio-video pre-training. Experimental validation on multiple benchmarks shows that our method achieves a 3.69%p of relative performance gain in zero-shot retrieval tasks compared to strong continual learning baselines, while reducing memory consumption by ~45%.
Authors: Evangelia Panourgia (Athens University of Economics,Business), Theodoros Plessas (Athens University of Economics,Business), Ilias Balampanis (Athens University of Economics,Business), Diomidis Spinellis (Athens University of Economics,Business, Delft University of Technology)
Abstract: The rising popularity of deep learning (DL) methods and techniques has invigorated interest in the topic of SE4DL (Software Engineering for Deep Learning), the application of software engineering (SE) practices on deep learning software. Despite the novel engineering challenges brought on by the data-driven and non-deterministic paradigm of DL software, little work has been invested into developing DL-targeted SE tools. On the other hand, tools tackling non-SE issues specific to DL are actively used and referred to under the umbrella term "MLOps (Machine Learning Operations) tools". Nevertheless, the available literature supports the utility of conventional SE tooling in DL software development. Building upon previous mining software repositories (MSR) research on tool usage in open-source software works, we identify conventional and MLOps tools adopted in popular applied DL projects that use Python as the main programming language. About 63\% of the GitHub repositories we examined contained at least one conventional SE tool. Software construction tools are the most widely adopted, while the opposite applies to management and maintenance tools. Relatively few MLOps tools were found to be use, with only 20 tools out of a sample of 74 used in at least one repository. The majority of them were open-source rather than proprietary. One of these tools, TensorBoard, was found to be adopted in about half of the repositories in our study. Consequently, the widespread use of conventional SE tooling demonstrates its relevance to DL software. Further research is recommended on the adoption of MLOps tooling, focusing on the relevance of particular tool types, the development of required tools, as well as ways to promote the use of already available tools.
Authors: Ishan Kumar, Zhijing Jin, Ehsan Mokhtarian, Siyuan Guo, Yuen Chen, Mrinmaya Sachan, Bernhard Sch\"olkopf
Abstract: Citation count of a paper is a commonly used proxy for evaluating the significance of a paper in the scientific community. Yet citation measures are widely criticized for failing to accurately reflect the true impact of a paper. Thus, we propose CausalCite, a new way to measure the significance of a paper by assessing the causal impact of the paper on its follow-up papers. CausalCite is based on a novel causal inference method, TextMatch, which adapts the traditional matching framework to high-dimensional text embeddings. TextMatch encodes each paper using text embeddings from large language models (LLMs), extracts similar samples by cosine similarity, and synthesizes a counterfactual sample as the weighted average of similar papers according to their similarity values. We demonstrate the effectiveness of CausalCite on various criteria, such as high correlation with paper impact as reported by scientific experts on a previous dataset of 1K papers, (test-of-time) awards for past papers, and its stability across various subfields of AI. We also provide a set of findings that can serve as suggested ways for future researchers to use our metric for a better understanding of the quality of a paper. Our code is available at https://github.com/causalNLP/causal-cite.
Authors: Xin Zhang, Jiawei Du, Yunsong Li, Weiying Xie, Joey Tianyi Zhou
Abstract: Dataset pruning aims to construct a coreset capable of achieving performance comparable to the original, full dataset. Most existing dataset pruning methods rely on snapshot-based criteria to identify representative samples, often resulting in poor generalization across various pruning and cross-architecture scenarios. Recent studies have addressed this issue by expanding the scope of training dynamics considered, including factors such as forgetting event and probability change, typically using an averaging approach. However, these works struggle to integrate a broader range of training dynamics without overlooking well-generalized samples, which may not be sufficiently highlighted in an averaging manner. In this study, we propose a novel dataset pruning method termed as Temporal Dual-Depth Scoring (TDDS), to tackle this problem. TDDS utilizes a dual-depth strategy to achieve a balance between incorporating extensive training dynamics and identifying representative samples for dataset pruning. In the first depth, we estimate the series of each sample's individual contributions spanning the training progress, ensuring comprehensive integration of training dynamics. In the second depth, we focus on the variability of the sample-wise contributions identified in the first depth to highlight well-generalized samples. Extensive experiments conducted on CIFAR and ImageNet datasets verify the superiority of TDDS over previous SOTA methods. Specifically on CIFAR-100, our method achieves 54.51% accuracy with only 10% training data, surpassing random selection by 7.83% and other comparison methods by at least 12.69%.
Authors: Michael Plainer, Hannes St\"ark, Charlotte Bunne, Stephan G\"unnemann
Abstract: Sampling all possible transition paths between two 3D states of a molecular system has various applications ranging from catalyst design to drug discovery. Current approaches to sample transition paths use Markov chain Monte Carlo and rely on time-intensive molecular dynamics simulations to find new paths. Our approach operates in the latent space of a normalizing flow that maps from the molecule's Boltzmann distribution to a Gaussian, where we propose new paths without requiring molecular simulations. Using alanine dipeptide, we explore Metropolis-Hastings acceptance criteria in the latent space for exact sampling and investigate different latent proposal mechanisms.
Authors: Sergei Shumilin, Alexander Ryabov, Serguei Barannikov, Evgeny Burnaev, Vladimir Vanovskii
Abstract: Voronoi tessellation, also known as Voronoi diagram, is an important computational geometry technique that has applications in various scientific disciplines. It involves dividing a given space into regions based on the proximity to a set of points. Autodifferentiation is a powerful tool for solving optimization tasks. Autodifferentiation assumes constructing a computational graph that allows to compute gradients using backpropagation algorithm. However, often the Voronoi tessellation remains the only non-differentiable part of a pipeline, prohibiting end-to-end differentiation. We present the method for autodifferentiation of the 2D Voronoi tessellation. The method allows one to construct the Voronoi tessellation and pass gradients, making the construction end-to-end differentiable. We provide the implementation details and present several important applications. To the best of our knowledge this is the first autodifferentiable realization of the Voronoi tessellation providing full set of Voronoi geometrical parameters in a differentiable way.
Authors: Jin-Long Wu, Matthew E. Levine, Tapio Schneider, Andrew Stuart
Abstract: Complex dynamical systems are notoriously difficult to model because some degrees of freedom (e.g., small scales) may be computationally unresolvable or are incompletely understood, yet they are dynamically important. For example, the small scales of cloud dynamics and droplet formation are crucial for controlling climate, yet are unresolvable in global climate models. Semi-empirical closure models for the effects of unresolved degrees of freedom often exist and encode important domain-specific knowledge. Building on such closure models and correcting them through learning the structural errors can be an effective way of fusing data with domain knowledge. Here we describe a general approach, principles, and algorithms for learning about structural errors. Key to our approach is to include structural error models inside the models of complex systems, for example, in closure models for unresolved scales. The structural errors then map, usually nonlinearly, to observable data. As a result, however, mismatches between model output and data are only indirectly informative about structural errors, due to a lack of labeled pairs of inputs and outputs of structural error models. Additionally, derivatives of the model may not exist or be readily available. We discuss how structural error models can be learned from indirect data with derivative-free Kalman inversion algorithms and variants, how sparsity constraints enforce a "do no harm" principle, and various ways of modeling structural errors. We also discuss the merits of using non-local and/or stochastic error models. In addition, we demonstrate how data assimilation techniques can assist the learning about structural errors in non-ergodic systems. The concepts and algorithms are illustrated in two numerical examples based on the Lorenz-96 system and a human glucose-insulin model.
Authors: Edward Sanderson, Bogdan J. Matuszewski
Abstract: Solutions to vision tasks in gastrointestinal endoscopy (GIE) conventionally use image encoders pretrained in a supervised manner with ImageNet-1k as backbones. However, the use of modern self-supervised pretraining algorithms and a recent dataset of 100k unlabelled GIE images (Hyperkvasir-unlabelled) may allow for improvements. In this work, we study the fine-tuned performance of models with ResNet50 and ViT-B backbones pretrained in self-supervised and supervised manners with ImageNet-1k and Hyperkvasir-unlabelled (self-supervised only) in a range of GIE vision tasks. In addition to identifying the most suitable pretraining pipeline and backbone architecture for each task, out of those considered, our results suggest three general principles. Firstly, that self-supervised pretraining generally produces more suitable backbones for GIE vision tasks than supervised pretraining. Secondly, that self-supervised pretraining with ImageNet-1k is typically more suitable than pretraining with Hyperkvasir-unlabelled, with the notable exception of monocular depth estimation in colonoscopy. Thirdly, that ViT-Bs are more suitable in polyp segmentation and monocular depth estimation in colonoscopy, ResNet50s are more suitable in polyp detection, and both architectures perform similarly in anatomical landmark recognition and pathological finding characterisation. We hope this work draws attention to the complexity of pretraining for GIE vision tasks, informs this development of more suitable approaches than the convention, and inspires further research on this topic to help advance this development. Code available: \underline{github.com/ESandML/SSL4GIE}
Authors: Nianzu Yang, Kaipeng Zeng, Haotian Lu, Yexin Wu, Zexin Yuan, Danni Chen, Shengdian Jiang, Jiaxiang Wu, Yimin Wang, Junchi Yan
Abstract: Neuronal morphology is essential for studying brain functioning and understanding neurodegenerative disorders. As acquiring real-world morphology data is expensive, computational approaches for morphology generation have been studied. Traditional methods heavily rely on expert-set rules and parameter tuning, making it difficult to generalize across different types of morphologies. Recently, MorphVAE was introduced as the sole learning-based method, but its generated morphologies lack plausibility, i.e., they do not appear realistic enough and most of the generated samples are topologically invalid. To fill this gap, this paper proposes MorphGrower, which mimicks the neuron natural growth mechanism for generation. Specifically, MorphGrower generates morphologies layer by layer, with each subsequent layer conditioned on the previously generated structure. During each layer generation, MorphGrower utilizes a pair of sibling branches as the basic generation block and generates branch pairs synchronously. This approach ensures topological validity and allows for fine-grained generation, thereby enhancing the realism of the final generated morphologies. Results on four real-world datasets demonstrate that MorphGrower outperforms MorphVAE by a notable margin. Importantly, the electrophysiological response simulation demonstrates the plausibility of our generated samples from a neuroscience perspective. Our code is available at https://github.com/Thinklab-SJTU/MorphGrower.
Authors: Ryan Abbott, Aleksandar Botev, Denis Boyda, Daniel C. Hackett, Gurtej Kanwar, S\'ebastien Racani\`ere, Danilo J. Rezende, Fernando Romero-L\'opez, Phiala E. Shanahan, Julian M. Urban
Abstract: Machine-learned normalizing flows can be used in the context of lattice quantum field theory to generate statistically correlated ensembles of lattice gauge fields at different action parameters. This work demonstrates how these correlations can be exploited for variance reduction in the computation of observables. Three different proof-of-concept applications are demonstrated using a novel residual flow architecture: continuum limits of gauge theories, the mass dependence of QCD observables, and hadronic matrix elements based on the Feynman-Hellmann approach. In all three cases, it is shown that statistical uncertainties are significantly reduced when machine-learned flows are incorporated as compared with the same calculations performed with uncorrelated ensembles or direct reweighting.
Authors: Blake Bordelon, Alexander Atanasov, Cengiz Pehlevan
Abstract: On a variety of tasks, the performance of neural networks predictably improves with training time, dataset size and model size across many orders of magnitude. This phenomenon is known as a neural scaling law. Of fundamental importance is the compute-optimal scaling law, which reports the performance as a function of units of compute when choosing model sizes optimally. We analyze a random feature model trained with gradient descent as a solvable model of network training and generalization. This reproduces many observations about neural scaling laws. First, our model makes a prediction about why the scaling of performance with training time and with model size have different power law exponents. Consequently, the theory predicts an asymmetric compute-optimal scaling rule where the number of training steps are increased faster than model parameters, consistent with recent empirical observations. Second, it has been observed that early in training, networks converge to their infinite-width dynamics at a rate $1/\textit{width}$ but at late time exhibit a rate $\textit{width}^{-c}$, where $c$ depends on the structure of the architecture and task. We show that our model exhibits this behavior. Lastly, our theory shows how the gap between training and test loss can gradually build up over time due to repeated reuse of data.
Authors: Anastasios N. Angelopoulos, Rina Foygel Barber, Stephen Bates
Abstract: We introduce a method for online conformal prediction with decaying step sizes. Like previous methods, ours possesses a retrospective guarantee of coverage for arbitrary sequences. However, unlike previous methods, we can simultaneously estimate a population quantile when it exists. Our theory and experiments indicate substantially improved practical properties: in particular, when the distribution is stable, the coverage is close to the desired level for every time point, not just on average over the observed sequence.
Authors: Zhifeng Kong, Arushi Goel, Rohan Badlani, Wei Ping, Rafael Valle, Bryan Catanzaro
Abstract: Augmenting large language models (LLMs) to understand audio -- including non-speech sounds and non-verbal speech -- is critically important for diverse real-world applications of LLMs. In this paper, we propose Audio Flamingo, a novel audio language model with 1) strong audio understanding abilities, 2) the ability to quickly adapt to unseen tasks via in-context learning and retrieval, and 3) strong multi-turn dialogue abilities. We introduce a series of training techniques, architecture design, and data strategies to enhance our model with these abilities. Extensive evaluations across various audio understanding tasks confirm the efficacy of our method, setting new state-of-the-art benchmarks. Our demo website is https://audioflamingo.github.io/ and the code is open-sourced at https://github.com/NVIDIA/audio-flamingo.
URLs: https://audioflamingo.github.io/, https://github.com/NVIDIA/audio-flamingo.
Authors: Bumsoo Kim, Abdul Muqeet, Kyuchul Lee, Sanghyun Seo
Abstract: Face re-aging is a prominent field in computer vision and graphics, with significant applications in photorealistic domains such as movies, advertising, and live streaming. Recently, the need to apply face re-aging to non-photorealistic images, like comics, illustrations, and animations, has emerged as an extension in various entertainment sectors. However, the lack of a network that can seamlessly edit the apparent age in NPR images has limited these tasks to a naive, sequential approach. This often results in unpleasant artifacts and a loss of facial attributes due to domain discrepancies. In this paper, we introduce a novel one-stage method for face re-aging combined with portrait style transfer, executed in a single generative step. We leverage existing face re-aging and style transfer networks, both trained within the same PR domain. Our method uniquely fuses distinct latent vectors, each responsible for managing aging-related attributes and NPR appearance. By adopting an exemplar-based approach, our method offers greater flexibility compared to domain-level fine-tuning approaches, which typically require separate training or fine-tuning for each domain. This effectively addresses the limitation of requiring paired datasets for re-aging and domain-level, data-driven approaches for stylization. Our experiments show that our model can effortlessly generate re-aged images while simultaneously transferring the style of examples, maintaining both natural appearance and controllability.
Authors: Chaojun Xiao, Pengle Zhang, Xu Han, Guangxuan Xiao, Yankai Lin, Zhengyan Zhang, Zhiyuan Liu, Maosong Sun
Abstract: Large language models (LLMs) have emerged as a cornerstone in real-world applications with lengthy streaming inputs (e.g., LLM-driven agents). However, existing LLMs, pre-trained on sequences with a restricted maximum length, cannot process longer sequences due to the out-of-domain and distraction issues. Common solutions often involve continual pre-training on longer sequences, which will introduce expensive computational overhead and uncontrollable change in model capabilities. In this paper, we unveil the intrinsic capacity of LLMs for understanding extremely long sequences without any fine-tuning. To this end, we introduce a training-free memory-based method, InfLLM. Specifically, InfLLM stores distant contexts into additional memory units and employs an efficient mechanism to lookup token-relevant units for attention computation. Thereby, InfLLM allows LLMs to efficiently process long sequences with a limited context window and well capture long-distance dependencies. Without any training, InfLLM enables LLMs that are pre-trained on sequences consisting of a few thousand tokens to achieve comparable performance with competitive baselines that continually train these LLMs on long sequences. Even when the sequence length is scaled to $1,024$K, InfLLM still effectively captures long-distance dependencies. Our code can be found in \url{https://github.com/thunlp/InfLLM}.
Authors: Darshana Saravanan, Naresh Manwani, Vineet Gandhi
Abstract: Partial label learning (PLL) is a weakly-supervised learning paradigm where each training instance is paired with a set of candidate labels (partial label), one of which is the true label. Noisy PLL (NPLL) relaxes this constraint by allowing some partial labels to not contain the true label, enhancing the practicality of the problem. Our work centres on NPLL and presents a minimalistic framework that initially assigns pseudo-labels to images by exploiting the noisy partial labels through a weighted nearest neighbour algorithm. These pseudo-label and image pairs are then used to train a deep neural network classifier with label smoothing. The classifier's features and predictions are subsequently employed to refine and enhance the accuracy of pseudo-labels. We perform thorough experiments on seven datasets and compare against nine NPLL and PLL methods. We achieve state-of-the-art results in all studied settings from the prior literature, obtaining substantial gains in fine-grained classification and extreme noise scenarios. Further, we show the promising generalisation capability of our framework in realistic crowd-sourced datasets.
Authors: Kang Du, Yu Xiang
Abstract: We study the data-generating mechanism for reconstructive SSL to shed light on its effectiveness. With an infinite amount of labeled samples, we provide a sufficient and necessary condition for perfect linear approximation. The condition reveals a full-rank component that preserves the label classes of Y, along with a redundant component. Motivated by the condition, we propose to approximate the redundant component by a low-rank factorization and measure the approximation quality by introducing a new quantity $\epsilon_s$, parameterized by the rank of factorization s. We incorporate $\epsilon_s$ into the excess risk analysis under both linear regression and ridge regression settings, where the latter regularization approach is to handle scenarios when the dimension of the learned features is much larger than the number of labeled samples n for downstream tasks. We design three stylized experiments to compare SSL with supervised learning under different settings to support our theoretical findings.
Authors: Razvan-Andrei Lascu, Mateusz B. Majka, {\L}ukasz Szpruch
Abstract: We study two variants of the mirror descent-ascent algorithm for solving min-max problems on the space of measures: simultaneous and sequential. We work under assumptions of convexity-concavity and relative smoothness of the payoff function with respect to a suitable Bregman divergence, defined on the space of measures via flat derivatives. We show that the convergence rates to mixed Nash equilibria, measured in the Nikaid\`o-Isoda error, are of order $\mathcal{O}\left(N^{-1/2}\right)$ and $\mathcal{O}\left(N^{-2/3}\right)$ for the simultaneous and sequential schemes, respectively, which is in line with the state-of-the-art results for related finite-dimensional algorithms.
Authors: Wilbert Pumacay, Ishika Singh, Jiafei Duan, Ranjay Krishna, Jesse Thomason, Dieter Fox
Abstract: To realize effective large-scale, real-world robotic applications, we must evaluate how well our robot policies adapt to changes in environmental conditions. Unfortunately, a majority of studies evaluate robot performance in environments closely resembling or even identical to the training setup. We present THE COLOSSEUM, a novel simulation benchmark, with 20 diverse manipulation tasks, that enables systematical evaluation of models across 14 axes of environmental perturbations. These perturbations include changes in color, texture, and size of objects, table-tops, and backgrounds; we also vary lighting, distractors, physical properties perturbations and camera pose. Using THE COLOSSEUM, we compare 5 state-of-the-art manipulation models to reveal that their success rate degrades between 30-50% across these perturbation factors. When multiple perturbations are applied in unison, the success rate degrades $\geq$75%. We identify that changing the number of distractor objects, target object color, or lighting conditions are the perturbations that reduce model performance the most. To verify the ecological validity of our results, we show that our results in simulation are correlated ($\bar{R}^2 = 0.614$) to similar perturbations in real-world experiments. We open source code for others to use THE COLOSSEUM, and also release code to 3D print the objects used to replicate the real-world perturbations. Ultimately, we hope that THE COLOSSEUM will serve as a benchmark to identify modeling decisions that systematically improve generalization for manipulation. See https://robot-colosseum.github.io/ for more details.
Authors: Haotian Sun, Yuchen Zhuang, Wei Wei, Chao Zhang, Bo Dai
Abstract: Adapting state-of-the-art Large Language Models (LLMs) like GPT-4 and Gemini for specific tasks is challenging. Due to the opacity in their parameters, embeddings, and even output probabilities, existing fine-tuning adaptation methods are inapplicable. Consequently, adapting these black-box LLMs is only possible through their API services, raising concerns about transparency, privacy, and cost. To address these challenges, we introduce BBox-Adapter, a novel lightweight adapter for black-box LLMs. BBox-Adapter distinguishes target and source domain data by treating target data as positive and source data as negative. It employs a ranking-based Noise Contrastive Estimation (NCE) loss to promote the likelihood of target domain data while penalizing that of the source domain. Furthermore, it features an online adaptation mechanism, which incorporates real-time positive data sampling from ground-truth, human, or AI feedback, coupled with negative data from previous adaptations. Extensive experiments demonstrate BBox-Adapter's effectiveness and cost efficiency. It improves model performance by up to 6.77% across diverse tasks and domains, while reducing training and inference costs by 31.30x and 1.84x, respectively.
Authors: Rui Zhang, Hongwei Li, Rui Wen, Wenbo Jiang, Yuan Zhang, Michael Backes, Yun Shen, Yang Zhang
Abstract: The increasing demand for customized Large Language Models (LLMs) has led to the development of solutions like GPTs. These solutions facilitate tailored LLM creation via natural language prompts without coding. However, the trustworthiness of third-party custom versions of LLMs remains an essential concern. In this paper, we propose the first instruction backdoor attacks against applications integrated with untrusted customized LLMs (e.g., GPTs). Specifically, these attacks embed the backdoor into the custom version of LLMs by designing prompts with backdoor instructions, outputting the attacker's desired result when inputs contain the pre-defined triggers. Our attack includes 3 levels of attacks: word-level, syntax-level, and semantic-level, which adopt different types of triggers with progressive stealthiness. We stress that our attacks do not require fine-tuning or any modification to the backend LLMs, adhering strictly to GPTs development guidelines. We conduct extensive experiments on 6 prominent LLMs and 5 benchmark text classification datasets. The results show that our instruction backdoor attacks achieve the desired attack performance without compromising utility. Additionally, we propose two defense strategies and demonstrate their effectiveness in reducing such attacks. Our findings highlight the vulnerability and the potential risks of LLM customization such as GPTs.
Authors: Yo Joong Choe, Aaditya Ramdas
Abstract: In anytime-valid sequential inference, it is known that any admissible procedure must be based on e-processes, which are composite generalizations of test martingales that quantify the accumulated evidence against a composite null hypothesis at any arbitrary stopping time. This paper studies methods for combining e-processes constructed using different information sets (filtrations) for the same null. Although e-processes constructed in the same filtration can be combined effortlessly (e.g., by averaging), e-processes constructed in different filtrations cannot, because their validity in a coarser filtration does not translate to validity in a finer filtration. This issue arises in exchangeability tests, independence tests, and tests for comparing forecasts with lags. We first establish that a class of functions called adjusters allows us to lift e-processes from a coarser filtration into any finer filtration. We then introduce a characterization theorem for adjusters, formalizing a sense in which using adjusters is necessary. There are two major implications. First, if we have a powerful e-process in a coarsened filtration, then we readily have a powerful e-process in the original filtration. Second, when we coarsen the filtration to construct an e-process, there is an asymptotically logarithmic cost of recovering anytime-validity in the original filtration.
Authors: Ajay Patel, Colin Raffel, Chris Callison-Burch
Abstract: Large language models (LLMs) have become a dominant and important tool for NLP researchers in a wide range of tasks. Today, many researchers use LLMs in synthetic data generation, task evaluation, fine-tuning, distillation, and other model-in-the-loop research workflows. However, challenges arise when using these models that stem from their scale, their closed source nature, and the lack of standardized tooling for these new and emerging workflows. The rapid rise to prominence of these models and these unique challenges has had immediate adverse impacts on open science and on the reproducibility of work that uses them. In this paper, we introduce DataDreamer, an open source Python library that allows researchers to write simple code to implement powerful LLM workflows. DataDreamer also helps researchers adhere to best practices that we propose to encourage open science and reproducibility. The library and documentation are available at https://github.com/datadreamer-dev/DataDreamer .
Authors: Louis Grenioux, Maxence Noble, Marylou Gabri\'e, Alain Oliviero Durmus
Abstract: Building upon score-based learning, new interest in stochastic localization techniques has recently emerged. In these models, one seeks to noise a sample from the data distribution through a stochastic process, called observation process, and progressively learns a denoiser associated to this dynamics. Apart from specific applications, the use of stochastic localization for the problem of sampling from an unnormalized target density has not been explored extensively. This work contributes to fill this gap. We consider a general stochastic localization framework and introduce an explicit class of observation processes, associated with flexible denoising schedules. We provide a complete methodology, $\textit{Stochastic Localization via Iterative Posterior Sampling}$ (SLIPS), to obtain approximate samples of this dynamics, and as a by-product, samples from the target distribution. Our scheme is based on a Markov chain Monte Carlo estimation of the denoiser and comes with detailed practical guidelines. We illustrate the benefits and applicability of SLIPS on several benchmarks of multi-modal distributions, including Gaussian mixtures in increasing dimensions, Bayesian logistic regression and a high-dimensional field system from statistical-mechanics.
Authors: Yueqin Yin, Zhendong Wang, Yi Gu, Hai Huang, Weizhu Chen, Mingyuan Zhou
Abstract: In the field of large language models (LLMs), aligning models with the diverse preferences of users is a critical challenge. Direct Preference Optimization (DPO) has played a key role in this area. It works by using pairs of preferences derived from the same prompts, and it functions without needing an additional reward model. However, DPO does not fully reflect the complex nature of human learning, which often involves understanding contrasting responses to not only identical but also similar questions. To overcome this shortfall, we propose Relative Preference Optimization (RPO). RPO is designed to discern between more and less preferred responses derived from both identical and related prompts. It introduces a contrastive weighting mechanism, enabling the tuning of LLMs using a broader range of preference data, including both paired and unpaired sets. This approach expands the learning capabilities of the model, allowing it to leverage insights from a more varied set of prompts. Through empirical tests, including dialogue and summarization tasks, and evaluations using the AlpacaEval2.0 leaderboard, RPO has demonstrated a superior ability to align LLMs with user preferences and to improve their adaptability during the training process. Our code can be viewed at https://github.com/yinyueqin/relative-preference-optimization
URLs: https://github.com/yinyueqin/relative-preference-optimization
Authors: Liwen Sun, Abhineet Agarwal, Aaron Kornblith, Bin Yu, Chenyan Xiong
Abstract: In the emergency department (ED), patients undergo triage and multiple laboratory tests before diagnosis. This time-consuming process causes ED crowding which impacts patient mortality, medical errors, staff burnout, etc. This work proposes (time) cost-effective diagnostic assistance that leverages artificial intelligence systems to help ED clinicians make efficient and accurate diagnoses. In collaboration with ED clinicians, we use public patient data to curate MIMIC-ED-Assist, a benchmark for AI systems to suggest laboratory tests that minimize wait time while accurately predicting critical outcomes such as death. With MIMIC-ED-Assist, we develop ED-Copilot which sequentially suggests patient-specific laboratory tests and makes diagnostic predictions. ED-Copilot employs a pre-trained bio-medical language model to encode patient information and uses reinforcement learning to minimize ED wait time and maximize prediction accuracy. On MIMIC-ED-Assist, ED-Copilot improves prediction accuracy over baselines while halving average wait time from four hours to two hours. ED-Copilot can also effectively personalize treatment recommendations based on patient severity, further highlighting its potential as a diagnostic assistant. Since MIMIC-ED-Assist is a retrospective benchmark, ED-Copilot is restricted to recommend only observed tests. We show ED-Copilot achieves competitive performance without this restriction as the maximum allowed time increases. Our code is available at https://github.com/cxcscmu/ED-Copilot.
Authors: Zicheng Lin, Zhibin Gou, Tian Liang, Ruilin Luo, Haowei Liu, Yujiu Yang
Abstract: The ability of Large Language Models (LLMs) to critique and refine their reasoning is crucial for their application in evaluation, feedback provision, and self-improvement. This paper introduces CriticBench, a comprehensive benchmark designed to assess LLMs' abilities to critique and rectify their reasoning across a variety of tasks. CriticBench encompasses five reasoning domains: mathematical, commonsense, symbolic, coding, and algorithmic. It compiles 15 datasets and incorporates responses from three LLM families. Utilizing CriticBench, we evaluate and dissect the performance of 17 LLMs in generation, critique, and correction reasoning, i.e., GQC reasoning. Our findings reveal: (1) a linear relationship in GQC capabilities, with critique-focused training markedly enhancing performance; (2) a task-dependent variation in correction effectiveness, with logic-oriented tasks being more amenable to correction; (3) GQC knowledge inconsistencies that decrease as model size increases; and (4) an intriguing inter-model critiquing dynamic, where stronger models are better at critiquing weaker ones, while weaker models can surprisingly surpass stronger ones in their self-critique. We hope these insights into the nuanced critique-correct reasoning of LLMs will foster further research in LLM critique and self-improvement.
Authors: Yanjie Ze, Gu Zhang, Kangning Zhang, Chenyuan Hu, Muhan Wang, Huazhe Xu
Abstract: Imitation learning provides an efficient way to teach robots dexterous skills; however, learning complex skills robustly and generalizablely usually consumes large amounts of human demonstrations. To tackle this challenging problem, we present 3D Diffusion Policy (DP3), a novel visual imitation learning approach that incorporates the power of 3D visual representations into diffusion policies, a class of conditional action generative models. The core design of DP3 is the utilization of a compact 3D visual representation, extracted from sparse point clouds with an efficient point encoder. In our experiments involving 72 simulation tasks, DP3 successfully handles most tasks with just 10 demonstrations and surpasses baselines with a 24.2% relative improvement. In 4 real robot tasks, DP3 demonstrates precise control with a high success rate of 85%, given only 40 demonstrations of each task, and shows excellent generalization abilities in diverse aspects, including space, viewpoint, appearance, and instance. Interestingly, in real robot experiments, DP3 rarely violates safety requirements, in contrast to baseline methods which frequently do, necessitating human intervention. Our extensive evaluation highlights the critical importance of 3D representations in real-world robot learning. Videos, code, and data are available on https://3d-diffusion-policy.github.io .
Authors: Xin Wang, Yu Zheng, Zhongwei Wan, Mi Zhang
Abstract: The advancements in Large Language Models (LLMs) have been hindered by their substantial sizes, which necessitate LLM compression methods for practical deployment. Singular Value Decomposition (SVD) offers a promising solution for LLM compression. However, state-of-the-art SVD-based LLM compression methods have two key limitations: truncating smaller singular values may lead to higher compression loss, and the lack of update on the compressed weight after SVD truncation. In this work, we propose SVD-LLM, a new SVD-based LLM compression method that addresses the limitations of existing methods. SVD-LLM incorporates a truncation-aware data whitening strategy to ensure a direct mapping between singular values and compression loss. Moreover, SVD-LLM adopts a layer-wise closed-form model parameter update strategy to compensate for accuracy degradation under high compression ratios. We evaluate SVD-LLM on a total of 10 datasets and eight models from three different LLM families at four different scales. Our results demonstrate the superiority of SVD-LLM over state-of-the-arts, especially at high model compression ratios.
Authors: Xuanlei Zhao, Shenggan Cheng, Chang Chen, Zangwei Zheng, Ziming Liu, Zheming Yang, Yang You
Abstract: Scaling multi-dimensional transformers to long sequences is indispensable across various domains. However, the challenges of large memory requirements and slow speeds of such sequences necessitate sequence parallelism. All existing approaches fall under the category of embedded sequence parallelism, which are limited to shard along a single sequence dimension, thereby introducing significant communication overhead. However, the nature of multi-dimensional transformers involves independent calculations across multiple sequence dimensions. To this end, we propose Dynamic Sequence Parallelism (DSP) as a novel abstraction of sequence parallelism. DSP dynamically switches the parallel dimension among all sequences according to the computation stage with efficient resharding strategy. DSP offers significant reductions in communication costs, adaptability across modules, and ease of implementation with minimal constraints. Experimental evaluations demonstrate DSP's superiority over state-of-the-art embedded sequence parallelism methods by remarkable throughput improvements ranging from 32.2% to 10x, with less than 25% communication volume.
Authors: Mengru Wang, Ningyu Zhang, Ziwen Xu, Zekun Xi, Shumin Deng, Yunzhi Yao, Qishen Zhang, Linyi Yang, Jindong Wang, Huajun Chen
Abstract: This paper investigates using knowledge editing techniques to detoxify Large Language Models (LLMs). We construct a benchmark, SafeEdit, which covers nine unsafe categories with various powerful attack prompts and equips comprehensive metrics for systematic evaluation. We conduct experiments with several knowledge editing approaches, indicating that knowledge editing has the potential to detoxify LLMs with a limited impact on general performance efficiently. Then, we propose a simple yet effective baseline, dubbed Detoxifying with Intraoperative Neural Monitoring (DINM), to diminish the toxicity of LLMs within a few tuning steps via only one instance. We further provide an in-depth analysis of the internal mechanism for various detoxifying approaches, demonstrating that previous methods like SFT and DPO may merely suppress the activations of toxic parameters, while DINM mitigates the toxicity of the toxic parameters to a certain extent, making permanent adjustments. We hope that these insights could shed light on future work of developing detoxifying approaches and the underlying knowledge mechanisms of LLMs. Code and benchmark are available at https://github.com/zjunlp/EasyEdit.
Authors: Yiwei Chen, Chao Tang, Amir Aghabiglou, Chung San Chu, Yves Wiaux
Abstract: We propose a new approach for non-Cartesian magnetic resonance image reconstruction. While unrolled architectures provide robustness via data-consistency layers, embedding measurement operators in Deep Neural Network (DNN) can become impractical at large scale. Alternative Plug-and-Play (PnP) approaches, where the denoising DNNs are blind to the measurement setting, are not affected by this limitation and have also proven effective, but their highly iterative nature also affects scalability. To address this scalability challenge, we leverage the "Residual-to-Residual DNN series for high-Dynamic range imaging (R2D2)" approach recently introduced in astronomical imaging. R2D2's reconstruction is formed as a series of residual images, iteratively estimated as outputs of DNNs taking the previous iteration's image estimate and associated data residual as inputs. The method can be interpreted as a learned version of the Matching Pursuit algorithm. We demonstrate R2D2 in simulation, considering radial k-space sampling acquisition sequences. Our preliminary results suggest that R2D2 achieves: (i) suboptimal performance compared to its unrolled incarnation R2D2-Net, which is however non-scalable due to the necessary embedding of NUFFT-based data-consistency layers; (ii) superior reconstruction quality to a scalable version of R2D2-Net embedding an FFT-based approximation for data consistency; (iii) superior reconstruction quality to PnP, while only requiring few iterations.
Authors: Chen Wang, Jin Zhao, Jiaqi Gong
Abstract: Recent advancements in Large Language Models (LLMs), particularly those built on Transformer architectures, have significantly broadened the scope of natural language processing (NLP) applications, transcending their initial use in chatbot technology. This paper investigates the multifaceted applications of these models, with an emphasis on the GPT series. This exploration focuses on the transformative impact of artificial intelligence (AI) driven tools in revolutionizing traditional tasks like coding and problem-solving, while also paving new paths in research and development across diverse industries. From code interpretation and image captioning to facilitating the construction of interactive systems and advancing computational domains, Transformer models exemplify a synergy of deep learning, data analysis, and neural network design. This survey provides an in-depth look at the latest research in Transformer models, highlighting their versatility and the potential they hold for transforming diverse application sectors, thereby offering readers a comprehensive understanding of the current and future landscape of Transformer-based LLMs in practical applications.
Authors: Jumman Hossain, Abu-Zaher Faridee, Nirmalya Roy, Jade Freeman, Timothy Gregory, Theron T. Trout
Abstract: Autonomous robots operating in complex environments face the critical challenge of identifying and utilizing environmental cover for covert navigation to minimize exposure to potential threats. We propose EnCoMP, an enhanced navigation framework that integrates offline reinforcement learning and our novel Adaptive Threat-Aware Visibility Estimation (ATAVE) algorithm to enable robots to navigate covertly and efficiently in diverse outdoor settings. ATAVE is a dynamic probabilistic threat modeling technique that we designed to continuously assess and mitigate potential threats in real-time, enhancing the robot's ability to navigate covertly by adapting to evolving environmental and threat conditions. Moreover, our approach generates high-fidelity multi-map representations, including cover maps, potential threat maps, height maps, and goal maps from LiDAR point clouds, providing a comprehensive understanding of the environment. These multi-maps offer detailed environmental insights, helping in strategic navigation decisions. The goal map encodes the relative distance and direction to the target location, guiding the robot's navigation. We train a Conservative Q-Learning (CQL) model on a large-scale dataset collected from real-world environments, learning a robust policy that maximizes cover utilization, minimizes threat exposure, and maintains efficient navigation. We demonstrate our method's capabilities on a physical Jackal robot, showing extensive experiments across diverse terrains. These experiments demonstrate EnCoMP's superior performance compared to state-of-the-art methods, achieving a 95% success rate, 85% cover utilization, and reducing threat exposure to 10.5%, while significantly outperforming baselines in navigation efficiency and robustness.
Authors: Hugo Caselles-Dupr\'e, Charles Mellerio, Paul H\'erent, Aliz\'ee Lopez-Persem, Benoit B\'eranger, Mathieu Soularue, Pierre Fautrel, Gauthier Vernier, Matthieu Cord
Abstract: The reconstruction of images observed by subjects from fMRI data collected during visual stimuli has made strong progress in the past decade, thanks to the availability of extensive fMRI datasets and advancements in generative models for image generation. However, the application of visual reconstruction has remained limited. Reconstructing visual imagination presents a greater challenge, with potentially revolutionary applications ranging from aiding individuals with disabilities to verifying witness accounts in court. The primary hurdles in this field are the absence of data collection protocols for visual imagery and the lack of datasets on the subject. Traditionally, fMRI-to-image relies on data collected from subjects exposed to visual stimuli, which poses issues for generating visual imagery based on the difference of brain activity between visual stimulation and visual imagery. For the first time, we have compiled a substantial dataset (around 6h of scans) on visual imagery along with a proposed data collection protocol. We then train a modified version of an fMRI-to-image model and demonstrate the feasibility of reconstructing images from two modes of imagination: from memory and from pure imagination. The resulting pipeline we call Mind-to-Image marks a step towards creating a technology that allow direct reconstruction of visual imagery.
Authors: Beomseok Kang, Harshit Kumar, Minah Lee, Biswadeep Chakraborty, Saibal Mukhopadhyay
Abstract: Locally interacting dynamical systems, such as epidemic spread, rumor propagation through crowd, and forest fire, exhibit complex global dynamics originated from local, relatively simple, and often stochastic interactions between dynamic elements. Their temporal evolution is often driven by transitions between a finite number of discrete states. Despite significant advancements in predictive modeling through deep learning, such interactions among many elements have rarely explored as a specific domain for predictive modeling. We present Attentive Recurrent Neural Cellular Automata (AR-NCA), to effectively discover unknown local state transition rules by associating the temporal information between neighboring cells in a permutation-invariant manner. AR-NCA exhibits the superior generalizability across various system configurations (i.e., spatial distribution of states), data efficiency and robustness in extremely data-limited scenarios even in the presence of stochastic interactions, and scalability through spatial dimension-independent prediction.
Authors: Amit Sharma, Teodor-Dumitru Ene, Kishor Kunal, Mingjie Liu, Zafar Hasan, Haoxing Ren
Abstract: This paper presents a comparative analysis of total cost of ownership (TCO) and performance between domain-adapted large language models (LLM) and state-of-the-art (SoTA) LLMs , with a particular emphasis on tasks related to coding assistance for chip design. We examine the TCO and performance metrics of a domain-adaptive LLM, ChipNeMo, against two leading LLMs, Claude 3 Opus and ChatGPT-4 Turbo, to assess their efficacy in chip design coding generation. Through a detailed evaluation of the accuracy of the model, training methodologies, and operational expenditures, this study aims to provide stakeholders with critical information to select the most economically viable and performance-efficient solutions for their specific needs. Our results underscore the benefits of employing domain-adapted models, such as ChipNeMo, that demonstrate improved performance at significantly reduced costs compared to their general-purpose counterparts. In particular, we reveal the potential of domain-adapted LLMs to decrease TCO by approximately 90%-95%, with the cost advantages becoming increasingly evident as the deployment scale expands. With expansion of deployment, the cost benefits of ChipNeMo become more pronounced, making domain-adaptive LLMs an attractive option for organizations with substantial coding needs supported by LLMs
Authors: Yanru Qu, Keyue Qiu, Yuxuan Song, Jingjing Gong, Jiawei Han, Mingyue Zheng, Hao Zhou, Wei-Ying Ma
Abstract: Generative models for structure-based drug design (SBDD) have shown promising results in recent years. Existing works mainly focus on how to generate molecules with higher binding affinity, ignoring the feasibility prerequisites for generated 3D poses and resulting in false positives. We conduct thorough studies on key factors of ill-conformational problems when applying autoregressive methods and diffusion to SBDD, including mode collapse and hybrid continuous-discrete space. In this paper, we introduce MolCRAFT, the first SBDD model that operates in the continuous parameter space, together with a novel noise reduced sampling strategy. Empirical results show that our model consistently achieves superior performance in binding affinity with more stable 3D structure, demonstrating our ability to accurately model interatomic interactions. To our best knowledge, MolCRAFT is the first to achieve reference-level Vina Scores (-6.59 kcal/mol) with comparable molecular size, outperforming other strong baselines by a wide margin (-0.84 kcal/mol). Code is available at https://github.com/AlgoMole/MolCRAFT.
Authors: Tengjun Huang
Abstract: With the rise of Visual and Language Pretraining (VLP), an increasing number of downstream tasks are adopting the paradigm of pretraining followed by fine-tuning. Although this paradigm has demonstrated potential in various multimodal downstream tasks, its implementation in the remote sensing domain encounters some obstacles. Specifically, the tendency for same-modality embeddings to cluster together impedes efficient transfer learning. To tackle this issue, we review the aim of multimodal transfer learning for downstream tasks from a unified perspective, and rethink the optimization process based on three distinct objectives. We propose "Harmonized Transfer Learning and Modality Alignment (HarMA)", a method that simultaneously satisfies task constraints, modality alignment, and single-modality uniform alignment, while minimizing training overhead through parameter-efficient fine-tuning. Remarkably, without the need for external data for training, HarMA achieves state-of-the-art performance in two popular multimodal retrieval tasks in the field of remote sensing. Our experiments reveal that HarMA achieves competitive and even superior performance to fully fine-tuned models with only minimal adjustable parameters. Due to its simplicity, HarMA can be integrated into almost all existing multimodal pretraining models. We hope this method can facilitate the efficient application of large models to a wide range of downstream tasks while significantly reducing the resource consumption. Code is available at https://github.com/seekerhuang/HarMA.
Authors: Zhangyong Liang, Huanhuan Gao, Ji Zhang
Abstract: Data constitute the foundational component of the data economy and its marketplaces. Efficient and fair data valuation has emerged as a topic of significant interest.\ Many approaches based on marginal contribution have shown promising results in various downstream tasks. However, they are well known to be computationally expensive as they require training a large number of utility functions, which are used to evaluate the usefulness or value of a given dataset for a specific purpose. As a result, it has been recognized as infeasible to apply these methods to a data marketplace involving large-scale datasets. Consequently, a critical issue arises: how can the re-training of the utility function be avoided? To address this issue, we propose a novel data valuation method from the perspective of optimal control, named the neural dynamic data valuation (NDDV). Our method has solid theoretical interpretations to accurately identify the data valuation via the sensitivity of the data optimal control state. In addition, we implement a data re-weighting strategy to capture the unique features of data points, ensuring fairness through the interaction between data points and the mean-field states. Notably, our method requires only training once to estimate the value of all data points, significantly improving the computational efficiency. We conduct comprehensive experiments using different datasets and tasks. The results demonstrate that the proposed NDDV method outperforms the existing state-of-the-art data valuation methods in accurately identifying data points with either high or low values and is more computationally efficient.
Authors: Xiaoqi Zhao, Youwei Pang, Wei Ji, Baicheng Sheng, Jiaming Zuo, Lihe Zhang, Huchuan Lu
Abstract: Different from the context-independent (CI) concepts such as human, car, and airplane, context-dependent (CD) concepts require higher visual understanding ability, such as camouflaged object and medical lesion. Despite the rapid advance of many CD understanding tasks in respective branches, the isolated evolution leads to their limited cross-domain generalisation and repetitive technique innovation. Since there is a strong coupling relationship between foreground and background context in CD tasks, existing methods require to train separate models in their focused domains. This restricts their real-world CD concept understanding towards artificial general intelligence (AGI). We propose a unified model with a single set of parameters, Spider, which only needs to be trained once. With the help of the proposed concept filter driven by the image-mask group prompt, Spider is able to understand and distinguish diverse strong context-dependent concepts to accurately capture the Prompter's intention. Without bells and whistles, Spider significantly outperforms the state-of-the-art specialized models in 8 different context-dependent segmentation tasks, including 4 natural scenes (salient, camouflaged, and transparent objects and shadow) and 4 medical lesions (COVID-19, polyp, breast, and skin lesion with color colonoscopy, CT, ultrasound, and dermoscopy modalities). Besides, Spider shows obvious advantages in continuous learning. It can easily complete the training of new tasks by fine-tuning parameters less than 1\% and bring a tolerable performance degradation of less than 5\% for all old tasks. The source code will be publicly available at \href{https://github.com/Xiaoqi-Zhao-DLUT/Spider-UniCDSeg}{Spider-UniCDSeg}.
Authors: Gerardo Duran-Martin, Matias Altamirano, Alexander Y. Shestopaloff, Leandro S\'anchez-Betancourt, Jeremias Knoblauch, Matt Jones, Fran\c{c}ois-Xavier Briol, Kevin Murphy
Abstract: We derive a novel, provably robust, and closed-form Bayesian update rule for online filtering in state-space models in the presence of outliers and misspecified measurement models. Our method combines generalised Bayesian inference with filtering methods such as the extended and ensemble Kalman filter. We use the former to show robustness and the latter to ensure computational efficiency in the case of nonlinear models. Our method matches or outperforms other robust filtering methods (such as those based on variational Bayes) at a much lower computational cost. We show this empirically on a range of filtering problems with outlier measurements, such as object tracking, state estimation in high-dimensional chaotic systems, and online learning of neural networks.
Authors: Zhufeng Li, Sandeep S Cranganore, Nicholas Youngblut, Niki Kilbertus
Abstract: Leveraging the vast genetic diversity within microbiomes offers unparalleled insights into complex phenotypes, yet the task of accurately predicting and understanding such traits from genomic data remains challenging. We propose a framework taking advantage of existing large models for gene vectorization to predict habitat specificity from entire microbial genome sequences. Based on our model, we develop attribution techniques to elucidate gene interaction effects that drive microbial adaptation to diverse environments. We train and validate our approach on a large dataset of high quality microbiome genomes from different habitats. We not only demonstrate solid predictive performance, but also how sequence-level information of entire genomes allows us to identify gene associations underlying complex phenotypes. Our attribution recovers known important interaction networks and proposes new candidates for experimental follow up.
Authors: Yicheng Li, Haobo Zhang, Qian Lin
Abstract: The saturation effect refers to the phenomenon that the kernel ridge regression (KRR) fails to achieve the information theoretical lower bound when the smoothness of the underground truth function exceeds certain level. The saturation effect has been widely observed in practices and a saturation lower bound of KRR has been conjectured for decades. In this paper, we provide a proof of this long-standing conjecture.
Authors: Zhen Huang, Amy Aumpansub
Abstract: Deep learning has been shown to be a promising tool in detecting software vulnerabilities. In this work, we train neural networks with program slices extracted from the source code of C/C++ programs to detect software vulnerabilities. The program slices capture the syntax and semantic characteristics of vulnerability-related program constructs, including API function call, array usage, pointer usage, and arithmetic expression. To achieve a strong prediction model for both vulnerable code and non-vulnerable code, we compare different types of training data, different optimizers, and different types of neural networks. Our result shows that combining different types of characteristics of source code and using a balanced number of vulnerable program slices and non-vulnerable program slices produce a balanced accuracy in predicting both vulnerable code and non-vulnerable code. Among different neural networks, BGRU with the ADAM optimizer performs the best in detecting software vulnerabilities with an accuracy of 92.49%.
Authors: Rudolf Herdt, Louisa Kinzel, Johann Georg Maa{\ss}, Marvin Walther, Henning Fr\"ohlich, Tim Schubert, Peter Maass, Christian Patrick Schaaf
Abstract: Rodents employ a broad spectrum of ultrasonic vocalizations (USVs) for social communication. As these vocalizations offer valuable insights into affective states, social interactions, and developmental stages of animals, various deep learning approaches have aimed to automate both the quantitative (detection) and qualitative (classification) analysis of USVs. Here, we present the first systematic evaluation of different types of neural networks for USV classification. We assessed various feedforward networks, including a custom-built, fully-connected network and convolutional neural network, different residual neural networks (ResNets), an EfficientNet, and a Vision Transformer (ViT). Paired with a refined, entropy-based detection algorithm (achieving recall of 94.9% and precision of 99.3%), the best architecture (achieving 86.79% accuracy) was integrated into a fully automated pipeline capable of analyzing extensive USV datasets with high reliability. Additionally, users can specify an individual minimum accuracy threshold based on their research needs. In this semi-automated setup, the pipeline selectively classifies calls with high pseudo-probability, leaving the rest for manual inspection. Our study focuses exclusively on neonatal USVs. As part of an ongoing phenotyping study, our pipeline has proven to be a valuable tool for identifying key differences in USVs produced by mice with autism-like behaviors.
Authors: Cristian Bodnar, Wessel P. Bruinsma, Ana Lucic, Megan Stanley, Johannes Brandstetter, Patrick Garvan, Maik Riechert, Jonathan Weyn, Haiyu Dong, Anna Vaughan, Jayesh K. Gupta, Kit Tambiratnam, Alex Archibald, Elizabeth Heider, Max Welling, Richard E. Turner, Paris Perdikaris
Abstract: Deep learning foundation models are revolutionizing many facets of science by leveraging vast amounts of data to learn general-purpose representations that can be adapted to tackle diverse downstream tasks. Foundation models hold the promise to also transform our ability to model our planet and its subsystems by exploiting the vast expanse of Earth system data. Here we introduce Aurora, a large-scale foundation model of the atmosphere trained on over a million hours of diverse weather and climate data. Aurora leverages the strengths of the foundation modelling approach to produce operational forecasts for a wide variety of atmospheric prediction problems, including those with limited training data, heterogeneous variables, and extreme events. In under a minute, Aurora produces 5-day global air pollution predictions and 10-day high-resolution weather forecasts that outperform state-of-the-art classical simulation tools and the best specialized deep learning models. Taken together, these results indicate that foundation models can transform environmental forecasting.
Authors: Hongyi Pan, Emadeldeen Hamdan, Xin Zhu, Koushik Biswas, Ahmet Enis Cetin, Ulas Bagci
Abstract: Central to the Transformer architectures' effectiveness is the self-attention mechanism, a function that maps queries, keys, and values into a high-dimensional vector space. However, training the attention weights of queries, keys, and values is non-trivial from a state of random initialization. In this paper, we propose two methods. (i) We first address the initialization problem of Vision Transformers by introducing a simple, yet highly innovative, initialization approach utilizing Discrete Cosine Transform (DCT) coefficients. Our proposed DCT-based attention initialization marks a significant gain compared to traditional initialization strategies; offering a robust foundation for the attention mechanism. Our experiments reveal that the DCT-based initialization enhances the accuracy of Vision Transformers in classification tasks. (ii) We also recognize that since DCT effectively decorrelates image information in the frequency domain, this decorrelation is useful for compression because it allows the quantization step to discard many of the higher-frequency components. Based on this observation, we propose a novel DCT-based compression technique for the attention function of Vision Transformers. Since high-frequency DCT coefficients usually correspond to noise, we truncate the high-frequency DCT components of the input patches. Our DCT-based compression reduces the size of weight matrices for queries, keys, and values. While maintaining the same level of accuracy, our DCT compressed Swin Transformers obtain a considerable decrease in the computational overhead.
Authors: Abhishek Kumar, Sarfaroz Yunusov, Ali Emami
Abstract: Research on Large Language Models (LLMs) has often neglected subtle biases that, although less apparent, can significantly influence the models' outputs toward particular social narratives. This study addresses two such biases within LLMs: representative bias, which denotes a tendency of LLMs to generate outputs that mirror the experiences of certain identity groups, and affinity bias, reflecting the models' evaluative preferences for specific narratives or viewpoints. We introduce two novel metrics to measure these biases: the Representative Bias Score (RBS) and the Affinity Bias Score (ABS), and present the Creativity-Oriented Generation Suite (CoGS), a collection of open-ended tasks such as short story writing and poetry composition, designed with customized rubrics to detect these subtle biases. Our analysis uncovers marked representative biases in prominent LLMs, with a preference for identities associated with being white, straight, and men. Furthermore, our investigation of affinity bias reveals distinctive evaluative patterns within each model, akin to `bias fingerprints'. This trend is also seen in human evaluators, highlighting a complex interplay between human and machine bias perceptions.
Authors: Brendan Park, Madeline Janecek, Naser Ezzati-Jivan, Yifeng Li, Ali Emami
Abstract: Large Language Models (LLMs) have demonstrated remarkable success in tasks like the Winograd Schema Challenge (WSC), showcasing advanced textual common-sense reasoning. However, applying this reasoning to multimodal domains, where understanding text and images together is essential, remains a substantial challenge. To address this, we introduce WinoVis, a novel dataset specifically designed to probe text-to-image models on pronoun disambiguation within multimodal contexts. Utilizing GPT-4 for prompt generation and Diffusion Attentive Attribution Maps (DAAM) for heatmap analysis, we propose a novel evaluation framework that isolates the models' ability in pronoun disambiguation from other visual processing challenges. Evaluation of successive model versions reveals that, despite incremental advancements, Stable Diffusion 2.0 achieves a precision of 56.7% on WinoVis, only marginally surpassing random guessing. Further error analysis identifies important areas for future research aimed at advancing text-to-image models in their ability to interpret and interact with the complex visual world.
Authors: Yuhao Chen, Zhimu Wang, Bo Wen, Farhana Zulkernine
Abstract: Unstructured text in medical notes and dialogues contains rich information. Recent advancements in Large Language Models (LLMs) have demonstrated superior performance in question answering and summarization tasks on unstructured text data, outperforming traditional text analysis approaches. However, there is a lack of scientific studies in the literature that methodically evaluate and report on the performance of different LLMs, specifically for domain-specific data such as medical chart notes. We propose an evaluation approach to analyze the performance of open-source LLMs such as Llama2 and Mistral for medical summarization tasks, using GPT-4 as an assessor. Our innovative approach to quantitative evaluation of LLMs can enable quality control, support the selection of effective LLMs for specific tasks, and advance knowledge discovery in digital health.
Authors: Cong Zhang, Derrick Goh Xin Deik, Dexun Li, Hao Zhang, Yong Liu
Abstract: The rapid advancement of neural language models has sparked a new surge of intelligent agent research. Unlike traditional agents, large language model-based agents (LLM agents) have emerged as a promising paradigm for achieving artificial general intelligence (AGI) due to their superior reasoning and generalization capabilities. Effective planning is crucial for the success of LLM agents in real-world tasks, making it a highly pursued topic in the community. Current planning methods typically translate tasks into executable action sequences. However, determining a feasible or optimal sequence for complex tasks at fine granularity, which often requires compositing long chains of heterogeneous actions, remains challenging. This paper introduces Meta-Task Planning (MTP), a zero-shot methodology for collaborative LLM-based multi-agent systems that simplifies complex task planning by decomposing it into a hierarchy of subordinate tasks, or meta-tasks. Each meta-task is then mapped into executable actions. MTP was assessed on two rigorous benchmarks, TravelPlanner and API-Bank. Notably, MTP achieved an average $\sim40\%$ success rate on TravelPlanner, significantly higher than the state-of-the-art (SOTA) baseline ($2.92\%$), and outperforming $LLM_{api}$-4 with ReAct on API-Bank by $\sim14\%$, showing the immense potential of integrating LLM with multi-agent systems.
Authors: Jianqiao Lu, Zhiyang Dou, Hongru Wang, Zeyu Cao, Jianbo Dai, Yingjia Wan, Yinya Huang, Zhijiang Guo
Abstract: In this work, we propose a novel method named \textbf{Auto}mated Process Labeling via \textbf{C}onfidence \textbf{V}ariation (\textbf{\textsc{AutoCV}}) to enhance the reasoning capabilities of large language models (LLMs) by automatically annotating the reasoning steps. Our approach begins by training a verification model on the correctness of final answers, enabling it to generate automatic process annotations. This verification model assigns a confidence score to each reasoning step, indicating the probability of arriving at the correct final answer from that point onward. We detect relative changes in the verification's confidence scores across reasoning steps to automatically annotate the reasoning process. This alleviates the need for numerous manual annotations or the high computational costs associated with model-induced annotation approaches. We experimentally validate that the confidence variations learned by the verification model trained on the final answer correctness can effectively identify errors in the reasoning steps. Subsequently, we demonstrate that the process annotations generated by \textsc{AutoCV} can improve the accuracy of the verification model in selecting the correct answer from multiple outputs generated by LLMs. Notably, we achieve substantial improvements across five datasets in mathematics and commonsense reasoning. The source code of \textsc{AutoCV} is available at \url{https://github.com/rookie-joe/AUTOCV}.
Authors: Shuijing Liu, Kaiwen Hong, Neeloy Chakraborty, Katherine Driggs-Campbell
Abstract: We investigate the feasibility of deploying reinforcement learning (RL) policies for constrained crowd navigation using a low-fidelity simulator. We introduce a representation of the dynamic environment, separating human and obstacle representations. Humans are represented through detected states, while obstacles are represented as computed point clouds based on maps and robot localization. This representation enables RL policies trained in a low-fidelity simulator to deploy in real world with a reduced sim2real gap. Additionally, we propose a spatio-temporal graph to model the interactions between agents and obstacles. Based on the graph, we use attention mechanisms to capture the robot-human, human-human, and human-obstacle interactions. Our method significantly improves navigation performance in both simulated and real-world environments. Video demonstrations can be found at https://sites.google.com/view/constrained-crowdnav/home.
URLs: https://sites.google.com/view/constrained-crowdnav/home.
Authors: Jaewoo Lee, Sujin Yun, Taeyoung Yun, Jinkyoo Park
Abstract: Offline Reinforcement Learning (Offline RL) presents challenges of learning effective decision-making policies from static datasets without any online interactions. Data augmentation techniques, such as noise injection and data synthesizing, aim to improve Q-function approximation by smoothing the learned state-action region. However, these methods often fall short of directly improving the quality of offline datasets, leading to suboptimal results. In response, we introduce \textbf{GTA}, Generative Trajectory Augmentation, a novel generative data augmentation approach designed to enrich offline data by augmenting trajectories to be both high-rewarding and dynamically plausible. GTA applies a diffusion model within the data augmentation framework. GTA partially noises original trajectories and then denoises them with classifier-free guidance via conditioning on amplified return value. Our results show that GTA, as a general data augmentation strategy, enhances the performance of widely used offline RL algorithms in both dense and sparse reward settings. Furthermore, we conduct a quality analysis of data augmented by GTA and demonstrate that GTA improves the quality of the data. Our code is available at https://github.com/Jaewoopudding/GTA
Authors: Andrea Angiuli, Jean-Pierre Fouque, Mathieu Lauri\`ere, Mengrui Zhang
Abstract: Mean Field Control Games (MFCG), introduced in [Angiuli et al., 2022a], represent competitive games between a large number of large collaborative groups of agents in the infinite limit of number and size of groups. In this paper, we prove the convergence of a three-timescale Reinforcement Q-Learning (RL) algorithm to solve MFCG in a model-free approach from the point of view of representative agents. Our analysis uses a Q-table for finite state and action spaces updated at each discrete time-step over an infinite horizon. In [Angiuli et al., 2023], we proved convergence of two-timescale algorithms for MFG and MFC separately highlighting the need to follow multiple population distributions in the MFC case. Here, we integrate this feature for MFCG as well as three rates of update decreasing to zero in the proper ratios. Our technique of proof uses a generalization to three timescales of the two-timescale analysis in [Borkar, 1997]. We give a simple example satisfying the various hypothesis made in the proof of convergence and illustrating the performance of the algorithm.