Authors: Ethan Baron, Daniel Hocevar, Zach Salehe
Abstract: We propose a foundation model for soccer, which is able to predict subsequent actions in a soccer match from a given input sequence of actions. As a proof of concept, we train a transformer architecture on three seasons of data from a professional soccer league. We quantitatively and qualitatively compare the performance of this transformer architecture to two baseline models: a Markov model and a multi-layer perceptron. Additionally, we discuss potential applications of our model. We provide an open-source implementation of our methods at https://github.com/danielhocevar/Foundation-Model-for-Soccer.
URLs: https://github.com/danielhocevar/Foundation-Model-for-Soccer.
Authors: Shubha R. Kharel, Prashansa Mukim, Piotr Maj, Grzegorz W. Deptuch, Shinjae Yoo, Yihui Ren, Soumyajit Mandal
Abstract: Extreme edge-AI systems, such as those in readout ASICs for radiation detection, must operate under stringent hardware constraints such as micron-level dimensions, sub-milliwatt power, and nanosecond-scale speed while providing clear accuracy advantages over traditional architectures. Finding ideal solutions means identifying optimal AI and ASIC design choices from a design space that has explosively expanded during the merger of these domains, creating non-trivial couplings which together act upon a small set of solutions as constraints tighten. It is impractical, if not impossible, to manually determine ideal choices among possibilities that easily exceed billions even in small-size problems. Existing methods to bridge this gap have leveraged theoretical understanding of hardware to f architecture search. However, the assumptions made in computing such theoretical metrics are too idealized to provide sufficient guidance during the difficult search for a practical implementation. Meanwhile, theoretical estimates for many other crucial metrics (like delay) do not even exist and are similarly variable, dependent on parameters of the process design kit (PDK). To address these challenges, we present a study that employs intelligent search using multi-objective Bayesian optimization, integrating both neural network search and ASIC synthesis in the loop. This approach provides reliable feedback on the collective impact of all cross-domain design choices. We showcase the effectiveness of our approach by finding several Pareto-optimal design choices for effective and efficient neural networks that perform real-time feature extraction from input pulses within the individual pixels of a readout ASIC.
Authors: Jaden Fiotto-Kaufman, Alexander R Loftus, Eric Todd, Jannik Brinkmann, Caden Juang, Koyena Pal, Can Rager, Aaron Mueller, Samuel Marks, Arnab Sen Sharma, Francesca Lucchetti, Michael Ripa, Adam Belfki, Nikhil Prakash, Sumeet Multani, Carla Brodley, Arjun Guha, Jonathan Bell, Byron Wallace, David Bau
Abstract: The enormous scale of state-of-the-art foundation models has limited their accessibility to scientists, because customized experiments at large model sizes require costly hardware and complex engineering that is impractical for most researchers. To alleviate these problems, we introduce NNsight, an open-source Python package with a simple, flexible API that can express interventions on any PyTorch model by building computation graphs. We also introduce NDIF, a collaborative research platform providing researchers access to foundation-scale LLMs via the NNsight API. Code, documentation, and tutorials are available at https://www.nnsight.net.
URLs: https://www.nnsight.net.
Authors: Orson Mengara
Abstract: Since the advent of generative artificial intelligence, every company and researcher has been rushing to develop their own generative models, whether commercial or not. Given the large number of users of these powerful new tools, there is currently no intrinsically verifiable way to explain from the ground up what happens when LLMs (large language models) learn. For example, those based on automatic speech recognition systems, which have to rely on huge and astronomical amounts of data collected from all over the web to produce fast and efficient results, In this article, we develop a backdoor attack called MarketBackFinal 2.0, based on acoustic data poisoning, MarketBackFinal 2.0 is mainly based on modern stock market models. In order to show the possible vulnerabilities of speech-based transformers that may rely on LLMs.
Authors: Andr\'e F. Cruz, Moritz Hardt, Celestine Mendler-D\"unner
Abstract: Current question-answering benchmarks predominantly focus on accuracy in realizable prediction tasks. Conditioned on a question and answer-key, does the most likely token match the ground truth? Such benchmarks necessarily fail to evaluate language models' ability to quantify outcome uncertainty. In this work, we focus on the use of language models as risk scores for unrealizable prediction tasks. We introduce folktexts, a software package to systematically generate risk scores using large language models, and evaluate them against benchmark prediction tasks. Specifically, the package derives natural language tasks from US Census data products, inspired by popular tabular data benchmarks. A flexible API allows for any task to be constructed out of 28 census features whose values are mapped to prompt-completion pairs. We demonstrate the utility of folktexts through a sweep of empirical insights on 16 recent large language models, inspecting risk scores, calibration curves, and diverse evaluation metrics. We find that zero-shot risk sores have high predictive signal while being widely miscalibrated: base models overestimate outcome uncertainty, while instruction-tuned models underestimate uncertainty and generate over-confident risk scores.
Authors: Pier Giuseppe Sessa, Robert Dadashi, L\'eonard Hussenot, Johan Ferret, Nino Vieillard, Alexandre Ram\'e, Bobak Shariari, Sarah Perrin, Abe Friesen, Geoffrey Cideron, Sertan Girgin, Piotr Stanczyk, Andrea Michi, Danila Sinopalnikov, Sabela Ramos, Am\'elie H\'eliou, Aliaksei Severyn, Matt Hoffman, Nikola Momchev, Olivier Bachem
Abstract: Reinforcement learning from human feedback (RLHF) is a key driver of quality and safety in state-of-the-art large language models. Yet, a surprisingly simple and strong inference-time strategy is Best-of-N sampling that selects the best generation among N candidates. In this paper, we propose Best-of-N Distillation (BOND), a novel RLHF algorithm that seeks to emulate Best-of-N but without its significant computational overhead at inference time. Specifically, BOND is a distribution matching algorithm that forces the distribution of generations from the policy to get closer to the Best-of-N distribution. We use the Jeffreys divergence (a linear combination of forward and backward KL) to balance between mode-covering and mode-seeking behavior, and derive an iterative formulation that utilizes a moving anchor for efficiency. We demonstrate the effectiveness of our approach and several design choices through experiments on abstractive summarization and Gemma models. Aligning Gemma policies with BOND outperforms other RLHF algorithms by improving results on several benchmarks.
Authors: Kamyab Karimi, Ali Ghodratnama, Reza Tavakkoli-Moghaddam
Abstract: Breast cancer is not preventable because of its unknown causes. However, its early diagnosis increases patients' recovery chances. Machine learning (ML) can be utilized to improve treatment outcomes in healthcare operations while diminishing costs and time. In this research, we suggest two novel feature selection (FS) methods based upon an imperialist competitive algorithm (ICA) and a bat algorithm (BA) and their combination with ML algorithms. This study aims to enhance diagnostic models' efficiency and present a comprehensive analysis to help clinical physicians make much more precise and reliable decisions than before. K-nearest neighbors, support vector machine, decision tree, Naive Bayes, AdaBoost, linear discriminant analysis, random forest, logistic regression, and artificial neural network are some of the methods employed. This paper applied a distinctive integration of evaluation measures and ML algorithms using the wrapper feature selection based on ICA (WFSIC) and BA (WFSB) separately. We compared two proposed approaches for the performance of the classifiers. Also, we compared our best diagnostic model with previous works reported in the literature survey. Experimentations were performed on the Wisconsin diagnostic breast cancer dataset. Results reveal that the proposed framework that uses the BA with an accuracy of 99.12\%, surpasses the framework using the ICA and most previous works. Additionally, the RF classifier in the approach of FS based on BA emerges as the best model and outperforms others regarding its criteria. Besides, the results illustrate the role of our techniques in reducing the dataset dimensions up to 90\% and increasing the performance of diagnostic models by over 99\%. Moreover, the result demonstrates that there are more critical features than the optimum dataset obtained by proposed FS approaches that have been selected by most ML models.
Authors: Yihang Yao, Zhepeng Cen, Wenhao Ding, Haohong Lin, Shiqi Liu, Tingnan Zhang, Wenhao Yu, Ding Zhao
Abstract: Offline safe reinforcement learning (RL) aims to train a policy that satisfies constraints using a pre-collected dataset. Most current methods struggle with the mismatch between imperfect demonstrations and the desired safe and rewarding performance. In this paper, we introduce OASIS (cOnditionAl diStributIon Shaping), a new paradigm in offline safe RL designed to overcome these critical limitations. OASIS utilizes a conditional diffusion model to synthesize offline datasets, thus shaping the data distribution toward a beneficial target domain. Our approach makes compliance with safety constraints through effective data utilization and regularization techniques to benefit offline safe RL training. Comprehensive evaluations on public benchmarks and varying datasets showcase OASIS's superiority in benefiting offline safe RL agents to achieve high-reward behavior while satisfying the safety constraints, outperforming established baselines. Furthermore, OASIS exhibits high data efficiency and robustness, making it suitable for real-world applications, particularly in tasks where safety is imperative and high-quality demonstrations are scarce.
Authors: Manish Marwah, Asad Narayanan, Stephen Jou, Martin Arlitt, Maria Pospelova
Abstract: The cost of errors related to machine learning classifiers, namely, false positives and false negatives, are not equal and are application dependent. For example, in cybersecurity applications, the cost of not detecting an attack is very different from marking a benign activity as an attack. Various design choices during machine learning model building, such as hyperparameter tuning and model selection, allow a data scientist to trade-off between these two errors. However, most of the commonly used metrics to evaluate model quality, such as $F_1$ score, which is defined in terms of model precision and recall, treat both these errors equally, making it difficult for users to optimize for the actual cost of these errors. In this paper, we propose a new cost-aware metric, $C_{score}$ based on precision and recall that can replace $F_1$ score for model evaluation and selection. It includes a cost ratio that takes into account the differing costs of handling false positives and false negatives. We derive and characterize the new cost metric, and compare it to $F_1$ score. Further, we use this metric for model thresholding for five cybersecurity related datasets for multiple cost ratios. The results show an average cost savings of 49%.
Authors: Frieda Rong, Max Kleiman-Weiner
Abstract: Social rewards shape human behavior. During development, a caregiver guides a learner's behavior towards culturally aligned goals and values. How do these behaviors persist and generalize when the caregiver is no longer present, and the learner must continue autonomously? Here, we propose a model of value internalization where social feedback trains an internal social reward (ISR) model that generates internal rewards when social rewards are unavailable. Through empirical simulations, we show that an ISR model prevents agents from unlearning socialized behaviors and enables generalization in out-of-distribution tasks. We characterize the implications of incomplete internalization, akin to "reward hacking" on the ISR. Additionally, we show that our model internalizes prosocial behavior in a multi-agent environment. Our work provides a foundation for understanding how humans acquire and generalize values and offers insights for aligning AI with human values.
Authors: Nora Agah, Javad Mohammadi, Alex Aved, David Ferris, Erika Ardiles Cruz, Philip Morrone
Abstract: As the complexities of Dynamic Data Driven Applications Systems increase, preserving their resilience becomes more challenging. For instance, maintaining power grid resilience is becoming increasingly complicated due to the growing number of stochastic variables (such as renewable outputs) and extreme weather events that add uncertainty to the grid. Current optimization methods have struggled to accommodate this rise in complexity. This has fueled the growing interest in data-driven methods used to operate the grid, leading to more vulnerability to cyberattacks. One such disruption that is commonly discussed is the adversarial disruption, where the intruder attempts to add a small perturbation to input data in order to "manipulate" the system operation. During the last few years, work on adversarial training and disruptions on the power system has gained popularity. In this paper, we will first review these applications, specifically on the most common types of adversarial disruptions: evasion and poisoning disruptions. Through this review, we highlight the gap between poisoning and evasion research when applied to the power grid. This is due to the underlying assumption that model training is secure, leading to evasion disruptions being the primary type of studied disruption. Finally, we will examine the impacts of data poisoning interventions and showcase how they can endanger power grid resilience.
Authors: Alejandro L. Garc\'ia Navarro, Nataliia Koneva, Alfonso S\'anchez-Maci\'an, Jos\'e Alberto Hern\'andez
Abstract: Python has gained widespread popularity in the fields of machine learning, artificial intelligence, and data engineering due to its effectiveness and extensive libraries. R, on its side, remains a dominant language for statistical analysis and visualization. However, certain libraries have become outdated, limiting their functionality and performance. Users can use Python's advanced machine learning and AI capabilities alongside R's robust statistical packages by combining these two programming languages. This paper explores using R's reticulate package to call Python from R, providing practical examples and highlighting scenarios where this integration enhances productivity and analytical capabilities. With a few hello-world code snippets, we demonstrate how to run Python's scikit-learn, pytorch and OpenAI gym libraries for building Machine Learning, Deep Learning, and Reinforcement Learning projects easily.
Authors: Shuya Feng, Meisam Mohammady, Hanbin Hong, Shenao Yan, Ashish Kundu, Binghui Wang, Yuan Hong
Abstract: Differentially private federated learning (DP-FL) is a promising technique for collaborative model training while ensuring provable privacy for clients. However, optimizing the tradeoff between privacy and accuracy remains a critical challenge. To our best knowledge, we propose the first DP-FL framework (namely UDP-FL), which universally harmonizes any randomization mechanism (e.g., an optimal one) with the Gaussian Moments Accountant (viz. DP-SGD) to significantly boost accuracy and convergence. Specifically, UDP-FL demonstrates enhanced model performance by mitigating the reliance on Gaussian noise. The key mediator variable in this transformation is the R\'enyi Differential Privacy notion, which is carefully used to harmonize privacy budgets. We also propose an innovative method to theoretically analyze the convergence for DP-FL (including our UDP-FL ) based on mode connectivity analysis. Moreover, we evaluate our UDP-FL through extensive experiments benchmarked against state-of-the-art (SOTA) methods, demonstrating superior performance on both privacy guarantees and model performance. Notably, UDP-FL exhibits substantial resilience against different inference attacks, indicating a significant advance in safeguarding sensitive data in federated learning environments.
Authors: Jiuxiang Gu, Yingyu Liang, Zhenmei Shi, Zhao Song, Yufa Zhou
Abstract: Cross-attention has become a fundamental module nowadays in many important artificial intelligence applications, e.g., retrieval-augmented generation (RAG), system prompt, guided stable diffusion, and many so on. Ensuring cross-attention privacy is crucial and urgently needed because its key and value matrices may contain sensitive information about companies and their users, many of which profit solely from their system prompts or RAG data. In this work, we design a novel differential privacy (DP) data structure to address the privacy security of cross-attention with a theoretical guarantee. In detail, let $n$ be the input token length of system prompt/RAG data, $d$ be the feature dimension, $0 < \alpha \le 1$ be the relative error parameter, $R$ be the maximum value of the query and key matrices, $R_w$ be the maximum value of the value matrix, and $r,s,\epsilon_s$ be parameters of polynomial kernel methods. Then, our data structure requires $\widetilde{O}(ndr^2)$ memory consumption with $\widetilde{O}(nr^2)$ initialization time complexity and $\widetilde{O}(\alpha^{-1} r^2)$ query time complexity for a single token query. In addition, our data structure can guarantee that the user query is $(\epsilon, \delta)$-DP with $\widetilde{O}(n^{-1} \epsilon^{-1} \alpha^{-1/2} R^{2s} R_w r^2)$ additive error and $n^{-1} (\alpha + \epsilon_s)$ relative error between our output and the true answer. Furthermore, our result is robust to adaptive queries in which users can intentionally attack the cross-attention system. To our knowledge, this is the first work to provide DP for cross-attention. We believe it can inspire more privacy algorithm design in large generative models (LGMs).
Authors: Beichen Zhang, Liang Li, Zheng-Jun Zha, Jiebo Luo, Qingming Huang
Abstract: Active learning (AL) is designed to construct a high-quality labeled dataset by iteratively selecting the most informative samples. Such sampling heavily relies on data representation, while recently pre-training is popular for robust feature learning. However, as pre-training utilizes low-level pretext tasks that lack annotation, directly using pre-trained representation in AL is inadequate for determining the sampling score. To address this problem, we propose a downstream-pretext domain knowledge traceback (DOKT) method that traces the data interactions of downstream knowledge and pre-training guidance for selecting diverse and instructive samples near the decision boundary. DOKT consists of a traceback diversity indicator and a domain-based uncertainty estimator. The diversity indicator constructs two feature spaces based on the pre-training pretext model and the downstream knowledge from annotation, by which it locates the neighbors of unlabeled data from the downstream space in the pretext space to explore the interaction of samples. With this mechanism, DOKT unifies the data relations of low-level and high-level representations to estimate traceback diversity. Next, in the uncertainty estimator, domain mixing is designed to enforce perceptual perturbing to unlabeled samples with similar visual patches in the pretext space. Then the divergence of perturbed samples is measured to estimate the domain uncertainty. As a result, DOKT selects the most diverse and important samples based on these two modules. The experiments conducted on ten datasets show that our model outperforms other state-of-the-art methods and generalizes well to various application scenarios such as semantic segmentation and image captioning.
Authors: Jayneel Vora, Nader Bouacida, Aditya Krishnan, Prasant Mohapatra
Abstract: We introduce FedDM, a novel training framework designed for the federated training of diffusion models. Our theoretical analysis establishes the convergence of diffusion models when trained in a federated setting, presenting the specific conditions under which this convergence is guaranteed. We propose a suite of training algorithms that leverage the U-Net architecture as the backbone for our diffusion models. These include a basic Federated Averaging variant, FedDM-vanilla, FedDM-prox to handle data heterogeneity among clients, and FedDM-quant, which incorporates a quantization module to reduce the model update size, thereby enhancing communication efficiency across the federated network. We evaluate our algorithms on FashionMNIST (28x28 resolution), CIFAR-10 (32x32 resolution), and CelebA (64x64 resolution) for DDPMs, as well as LSUN Church Outdoors (256x256 resolution) for LDMs, focusing exclusively on the imaging modality. Our evaluation results demonstrate that FedDM algorithms maintain high generation quality across image resolutions. At the same time, the use of quantized updates and proximal terms in the local training objective significantly enhances communication efficiency (up to 4x) and model convergence, particularly in non-IID data settings, at the cost of increased FID scores (up to 1.75x).
Authors: Yonghao Liu, Mengyu Li, Ximing Li, Lan Huang, Fausto Giunchiglia, Yanchun Liang, Xiaoyue Feng, Renchu Guan
Abstract: Node classification is an essential problem in graph learning. However, many models typically obtain unsatisfactory performance when applied to few-shot scenarios. Some studies have attempted to combine meta-learning with graph neural networks to solve few-shot node classification on graphs. Despite their promising performance, some limitations remain. First, they employ the node encoding mechanism of homophilic graphs to learn node embeddings, even in heterophilic graphs. Second, existing models based on meta-learning ignore the interference of randomness in the learning process. Third, they are trained using only limited labeled nodes within the specific task, without explicitly utilizing numerous unlabeled nodes. Finally, they treat almost all sampled tasks equally without customizing them for their uniqueness. To address these issues, we propose a novel framework for few-shot node classification called Meta-GPS++. Specifically, we first adopt an efficient method to learn discriminative node representations on homophilic and heterophilic graphs. Then, we leverage a prototype-based approach to initialize parameters and contrastive learning for regularizing the distribution of node embeddings. Moreover, we apply self-training to extract valuable information from unlabeled nodes. Additionally, we adopt S$^2$ (scaling & shifting) transformation to learn transferable knowledge from diverse tasks. The results on real-world datasets show the superiority of Meta-GPS++. Our code is available here.
Authors: Yunseon Choi, Sangmin Bae, Seonghyun Ban, Minchan Jeong, Chuheng Zhang, Lei Song, Li Zhao, Jiang Bian, Kee-Eung Kim
Abstract: With the advent of foundation models, prompt tuning has positioned itself as an important technique for directing model behaviors and eliciting desired responses. Prompt tuning regards selecting appropriate keywords included into the input, thereby adapting to the downstream task without adjusting or fine-tuning the model parameters. There is a wide range of work in prompt tuning, from approaches that directly harness the backpropagated gradient signals from the model, to those employing black-box optimization such as reinforcement learning (RL) methods. Our primary focus is on RLPrompt, which aims to find optimal prompt tokens leveraging soft Q-learning. While the results show promise, we have observed that the prompts frequently appear unnatural, which impedes their interpretability. We address this limitation by using sparse Tsallis entropy regularization, a principled approach to filtering out unlikely tokens from consideration. We extensively evaluate our approach across various tasks, including few-shot text classification, unsupervised text style transfer, and textual inversion from images. The results indicate a notable improvement over baselines, highlighting the efficacy of our approach in addressing the challenges of prompt tuning. Moreover, we show that the prompts discovered using our method are more natural and interpretable compared to those from other baselines.
Authors: Fan Zhao, You Chen
Abstract: Deep learning-based methods for Time Series Classification (TSC) typically utilize deep networks to extract features, which are then processed through a combination of a Fully Connected (FC) layer and a SoftMax function. However, we have observed the phenomenon of inter-class similarity and intra-class inconsistency in the datasets from the UCR archive and further analyzed how this phenomenon adversely affects the "FC+SoftMax" paradigm. To address the issue, we introduce ECR, which, for the first time to our knowledge, applies deep learning-based retrieval algorithm to the TSC problem and integrates classification and retrieval models. Experimental results on 112 UCR datasets demonstrate that ECR is state-of-the-art(sota) compared to existing deep learning-based methods. Furthermore, we have developed a more precise classifier, ECRTime, which is an ensemble of ECR. ECRTime surpasses the currently most accurate deep learning classifier, InceptionTime, in terms of accuracy, achieving this with reduced training time and comparable scalability.
Authors: Hoang Pham, The-Anh Ta, Anh Tran, Khoa D. Doan
Abstract: Recently, backdoor attacks have become an emerging threat to the security of machine learning models. From the adversary's perspective, the implanted backdoors should be resistant to defensive algorithms, but some recently proposed fine-tuning defenses can remove these backdoors with notable efficacy. This is mainly due to the catastrophic forgetting (CF) property of deep neural networks. This paper counters CF of backdoors by leveraging continual learning (CL) techniques. We begin by investigating the connectivity between a backdoored and fine-tuned model in the loss landscape. Our analysis confirms that fine-tuning defenses, especially the more advanced ones, can easily push a poisoned model out of the backdoor regions, making it forget all about the backdoors. Based on this finding, we re-formulate backdoor training through the lens of CL and propose a novel framework, named Sequential Backdoor Learning (SBL), that can generate resilient backdoors. This framework separates the backdoor poisoning process into two tasks: the first task learns a backdoored model, while the second task, based on the CL principles, moves it to a backdoored region resistant to fine-tuning. We additionally propose to seek flatter backdoor regions via a sharpness-aware minimizer in the framework, further strengthening the durability of the implanted backdoor. Finally, we demonstrate the effectiveness of our method through extensive empirical experiments on several benchmark datasets in the backdoor domain. The source code is available at https://github.com/mail-research/SBL-resilient-backdoors
URLs: https://github.com/mail-research/SBL-resilient-backdoors
Authors: Sumeyye Bas, Kiymet Kaya, Resul Tugay, Sule Gunduz Oguducu
Abstract: Graphs are crucial for representing interrelated data and aiding predictive modeling by capturing complex relationships. Achieving high-quality graph representation is important for identifying linked patterns, leading to improvements in Graph Neural Networks (GNNs) to better capture data structures. However, challenges such as data scarcity, high collection costs, and ethical concerns limit progress. As a result, generative models and data augmentation have become more and more popular. This study explores using generated graphs for data augmentation, comparing the performance of combining generated graphs with real graphs, and examining the effect of different quantities of generated graphs on graph classification tasks. The experiments show that balancing scalability and quality requires different generators based on graph size. Our results introduce a new approach to graph data augmentation, ensuring consistent labels and enhancing classification performance.
Authors: Thomas Souverain, Johnathan Nguyen, Nicolas Meric, Paul \'Egr\'e
Abstract: In this paper, we propose an experimental investigation of the problem of AI fairness in classification. We train an AI model and develop our own fairness package FairDream to detect inequalities and then to correct for them, using income prediction as a case study. Our experiments show that it is a property of FairDream to fulfill fairness objectives which are conditional on the ground truth (Equalized Odds), even when the algorithm is set the task of equalizing positives across groups (Demographic Parity). While this may be seen as an anomaly, we explain this property by comparing our approach with a closely related fairness method (GridSearch), which can enforce Demographic Parity at the expense of Equalized Odds. We grant that a fairness metric conditioned on true labels does not give a sufficient criterion to reach fairness, but we argue that it gives us at least a necessary condition to implement Demographic Parity cautiously. We also explain why neither Equal Calibration nor Equal Precision stand as relevant fairness criteria in classification. Addressing their limitations to warn the decision-maker for any disadvantaging rate, Equalized Odds avoids the peril of strict conservatism, while keeping away the utopia of a whole redistribution of resources through algorithms.
Authors: Lirong Wu, Yunfan Liu, Haitao Lin, Yufei Huang, Stan Z. Li
Abstract: To bridge the gaps between powerful Graph Neural Networks (GNNs) and lightweight Multi-Layer Perceptron (MLPs), GNN-to-MLP Knowledge Distillation (KD) proposes to distill knowledge from a well-trained teacher GNN into a student MLP. In this paper, we revisit the knowledge samples (nodes) in teacher GNNs from the perspective of hardness, and identify that hard sample distillation may be a major performance bottleneck of existing graph KD algorithms. The GNN-to-MLP KD involves two different types of hardness, one student-free knowledge hardness describing the inherent complexity of GNN knowledge, and the other student-dependent distillation hardness describing the difficulty of teacher-to-student distillation. However, most of the existing work focuses on only one of these aspects or regards them as one thing. This paper proposes a simple yet effective Hardness-aware GNN-to-MLP Distillation (HGMD) framework, which decouples the two hardnesses and estimates them using a non-parametric approach. Finally, two hardness-aware distillation schemes (i.e., HGMD-weight and HGMD-mixup) are further proposed to distill hardness-aware knowledge from teacher GNNs into the corresponding nodes of student MLPs. As non-parametric distillation, HGMD does not involve any additional learnable parameters beyond the student MLPs, but it still outperforms most of the state-of-the-art competitors. HGMD-mixup improves over the vanilla MLPs by 12.95% and outperforms its teacher GNNs by 2.48% averaged over seven real-world datasets.
Authors: Yanxi Chen, Yaliang Li, Bolin Ding, Jingren Zhou
Abstract: We initiate a formal investigation into the design and analysis of LLM-based algorithms, i.e. algorithms that contain one or multiple calls of large language models (LLMs) as sub-routines and critically rely on the capabilities of LLMs. While LLM-based algorithms, ranging from basic LLM calls with prompt engineering to complicated LLM-powered agent systems and compound AI systems, have achieved remarkable empirical success, the design and optimization of them have mostly relied on heuristics and trial-and-errors, which is largely due to a lack of formal and analytical study for these algorithms. To fill this gap, we start by identifying the computational-graph representation of LLM-based algorithms, the design principle of task decomposition, and some key abstractions, which then facilitate our formal analysis for the accuracy and efficiency of LLM-based algorithms, despite the black-box nature of LLMs. We further consider parallel decomposition for a case study, providing extensive analytical and empirical study for four concrete examples of this pattern. Our proposed framework holds promise for advancing LLM-based algorithms, by revealing the reasons behind curious empirical phenomena, guiding the choices of hyperparameters, predicting the empirical performance of algorithms, and inspiring new algorithm design. To promote further study of LLM-based algorithms, we release our source code at https://github.com/modelscope/agentscope/tree/main/examples/paper_llm_based_algorithm.
URLs: https://github.com/modelscope/agentscope/tree/main/examples/paper_llm_based_algorithm.
Authors: Shusen Ma, Yu Kang, Peng Bai, Yun-Bo Zhao
Abstract: In multivariate time-series forecasting (MTSF), extracting the temporal correlations of the input sequences is crucial. While popular Transformer-based predictive models can perform well, their quadratic computational complexity results in inefficiency and high overhead. The recently emerged Mamba, a selective state space model, has shown promising results in many fields due to its strong temporal feature extraction capabilities and linear computational complexity. However, due to the unilateral nature of Mamba, channel-independent predictive models based on Mamba cannot attend to the relationships among all variables in the manner of Transformer-based models. To address this issue, we combine fast-attention with Mamba to introduce a novel framework named FMamba for MTSF. Technically, we first extract the temporal features of the input variables through an embedding layer, then compute the dependencies among input variables via the fast-attention module. Subsequently, we use Mamba to selectively deal with the input features and further extract the temporal dependencies of the variables through the multi-layer perceptron block (MLP-block). Finally, FMamba obtains the predictive results through the projector, a linear layer. Experimental results on eight public datasets demonstrate that FMamba can achieve state-of-the-art performance while maintaining low computational overhead.
Authors: Ze Yu Zhang, Arun Verma, Finale Doshi-Velez, Bryan Kian Hsiang Low
Abstract: Large language models (LLMs) are widely used in decision-making, but their reliability, especially in critical tasks like healthcare, is not well-established. Therefore, understanding how LLMs reason and make decisions is crucial for their safe deployment. This paper investigates how the uncertainty of responses generated by LLMs relates to the information provided in the input prompt. Leveraging the insight that LLMs learn to infer latent concepts during pretraining, we propose a prompt-response concept model that explains how LLMs generate responses and helps understand the relationship between prompts and response uncertainty. We show that the uncertainty decreases as the prompt's informativeness increases, similar to epistemic uncertainty. Our detailed experimental results on real datasets validate our proposed model.
Authors: Abdullahi Saka, Ridwan Taiwo, Nurudeen Saka, Benjamin Oluleye, Jamiu Dauda, Lukman Akanbi
Abstract: Effective management of construction and demolition waste (C&DW) is crucial for sustainable development, as the industry accounts for 40% of the waste generated globally. The effectiveness of the C&DW management relies on the proper quantification of C&DW to be generated. Despite demolition activities having larger contributions to C&DW generation, extant studies have focused on construction waste. The few extant studies on demolition are often from the regional level perspective and provide no circularity insights. Thus, this study advances demolition quantification via Variable Modelling (VM) with Machine Learning (ML). The demolition dataset of 2280 projects were leveraged for the ML modelling, with XGBoost model emerging as the best (based on the Copeland algorithm), achieving R2 of 0.9977 and a Mean Absolute Error of 5.0910 on the testing dataset. Through the integration of the ML model with Building Information Modelling (BIM), the study developed a system for predicting quantities of recyclable and landfill materials from building demolitions. This provides detailed insights into the circularity of demolition waste and facilitates better planning and management. The SHapley Additive exPlanations (SHAP) method highlighted the implications of the features for demolition waste circularity. The study contributes to empirical studies on pre-demolition auditing at the project level and provides practical tools for implementation. Its findings would benefit stakeholders in driving a circular economy in the industry.
Authors: A. Verdone, A. Devoto, C. Sebastiani, J. Carmignani, M. D'Onofrio, S. Giagu, S. Scardapane, M. Panella
Abstract: The experiments at the Large Hadron Collider at CERN generate vast amounts of complex data from high-energy particle collisions. This data presents significant challenges due to its volume and complex reconstruction, necessitating the use of advanced analysis techniques for analysis. Recent advancements in deep learning, particularly Graph Neural Networks, have shown promising results in addressing the challenges but remain computationally expensive. The study presented in this paper uses a simulated particle collision dataset to integrate influence analysis inside the graph classification pipeline aiming at improving the accuracy and efficiency of collision event prediction tasks. By using a Graph Neural Network for initial training, we applied a gradient-based data influence method to identify influential training samples and then we refined the dataset by removing non-contributory elements: the model trained on this new reduced dataset can achieve good performances at a reduced computational cost. The method is completely agnostic to the specific influence method: different influence modalities can be easily integrated into our methodology. Moreover, by analyzing the discarded elements we can provide further insights about the event classification task. The novelty of integrating data attribution techniques together with Graph Neural Networks in high-energy physics tasks can offer a robust solution for managing large-scale data problems, capturing critical patterns, and maximizing accuracy across several high-data demand domains.
Authors: Tingting Ou, Marco Avella Medina, Rachel Cummings
Abstract: In this work we first show that the classical Thompson sampling algorithm for multi-arm bandits is differentially private as-is, without any modification. We provide per-round privacy guarantees as a function of problem parameters and show composition over $T$ rounds; since the algorithm is unchanged, existing $O(\sqrt{NT\log N})$ regret bounds still hold and there is no loss in performance due to privacy. We then show that simple modifications -- such as pre-pulling all arms a fixed number of times, increasing the sampling variance -- can provide tighter privacy guarantees. We again provide privacy guarantees that now depend on the new parameters introduced in the modification, which allows the analyst to tune the privacy guarantee as desired. We also provide a novel regret analysis for this new algorithm, and show how the new parameters also impact expected regret. Finally, we empirically validate and illustrate our theoretical findings in two parameter regimes and demonstrate that tuning the new parameters substantially improve the privacy-regret tradeoff.
Authors: Haoran Shen, Chen Zeng, Jiahui Wang, Qiao Wang
Abstract: It has been observed that even a small amount of noise introduced into the dataset can significantly degrade the performance of KAN. In this brief note, we aim to quantitatively evaluate the performance when noise is added to the dataset. We propose an oversampling technique combined with denoising to alleviate the impact of noise. Specifically, we employ kernel filtering based on diffusion maps for pre-filtering the noisy data for training KAN network. Our experiments show that while adding i.i.d. noise with any fixed SNR, when we increase the amount of training data by a factor of $r$, the test-loss (RMSE) of KANs will exhibit a performance trend like $\text{test-loss} \sim \mathcal{O}(r^{-\frac{1}{2}})$ as $r\to +\infty$. We conclude that applying both oversampling and filtering strategies can reduce the detrimental effects of noise. Nevertheless, determining the optimal variance for the kernel filtering process is challenging, and enhancing the volume of training data substantially increases the associated costs, because the training dataset needs to be expanded multiple times in comparison to the initial clean data. As a result, the noise present in the data ultimately diminishes the effectiveness of Kolmogorov-Arnold networks.
Authors: Marvin Seyfarth, Salman Ul Hassan Dar, Sandy Engelhardt
Abstract: Contemporary developments in generative AI are rapidly transforming the field of medical AI. These developments have been predominantly driven by the availability of large datasets and high computing power, which have facilitated a significant increase in model capacity. Despite their considerable potential, these models demand substantially high power, leading to high carbon dioxide (CO2) emissions. Given the harm such models are causing to the environment, there has been little focus on the carbon footprints of such models. This study analyzes carbon emissions from 2D and 3D latent diffusion models (LDMs) during training and data generation phases, revealing a surprising finding: the synthesis of large images contributes most significantly to these emissions. We assess different scenarios including model sizes, image dimensions, distributed training, and data generation steps. Our findings reveal substantial carbon emissions from these models, with training 2D and 3D models comparable to driving a car for 10 km and 90 km, respectively. The process of data generation is even more significant, with CO2 emissions equivalent to driving 160 km for 2D models and driving for up to 3345 km for 3D synthesis. Additionally, we found that the location of the experiment can increase carbon emissions by up to 94 times, and even the time of year can influence emissions by up to 50%. These figures are alarming, considering they represent only a single training and data generation phase for each model. Our results emphasize the urgent need for developing environmentally sustainable strategies in generative AI.
Authors: Yuening Li, Xiao Fu, Junbin Liu, Wing-Kin Ma
Abstract: This work proposes a variational inference (VI) framework for hyperspectral unmixing in the presence of endmember variability (HU-EV). An EV-accounted noisy linear mixture model (LMM) is considered, and the presence of outliers is also incorporated into the model. Following the marginalized maximum likelihood (MML) principle, a VI algorithmic structure is designed for probabilistic inference for HU-EV. Specifically, a patch-wise static endmember assumption is employed to exploit spatial smoothness and to try to overcome the ill-posed nature of the HU-EV problem. The design facilitates lightweight, continuous optimization-based updates under a variety of endmember priors. Some of the priors, such as the Beta prior, were previously used under computationally heavy, sampling-based probabilistic HU-EV methods. The effectiveness of the proposed framework is demonstrated through synthetic, semi-real, and real-data experiments.
Authors: Alexey Skrynnik, Anton Andreychuk, Anatolii Borzilov, Alexander Chernyavskiy, Konstantin Yakovlev, Aleksandr Panov
Abstract: Multi-agent reinforcement learning (MARL) has recently excelled in solving challenging cooperative and competitive multi-agent problems in various environments with, mostly, few agents and full observability. Moreover, a range of crucial robotics-related tasks, such as multi-robot navigation and obstacle avoidance, that have been conventionally approached with the classical non-learnable methods (e.g., heuristic search) is currently suggested to be solved by the learning-based or hybrid methods. Still, in this domain, it is hard, not to say impossible, to conduct a fair comparison between classical, learning-based, and hybrid approaches due to the lack of a unified framework that supports both learning and evaluation. To this end, we introduce POGEMA, a set of comprehensive tools that includes a fast environment for learning, a generator of problem instances, the collection of pre-defined ones, a visualization toolkit, and a benchmarking tool that allows automated evaluation. We introduce and specify an evaluation protocol defining a range of domain-related metrics computed on the basics of the primary evaluation indicators (such as success rate and path length), allowing a fair multi-fold comparison. The results of such a comparison, which involves a variety of state-of-the-art MARL, search-based, and hybrid methods, are presented.
Authors: Athina Sotiropoulou, David Alvarez-Melis
Abstract: Optimal Transport (OT) has recently emerged as a powerful framework for learning minimal-displacement maps between distributions. The predominant approach involves a neural parametrization of the Monge formulation of OT, typically assuming the same space for both distributions. However, the setting across ``incomparable spaces'' (e.g., of different dimensionality), corresponding to the Gromov- Wasserstein distance, remains underexplored, with existing methods often imposing restrictive assumptions on the cost function. In this paper, we present a novel neural formulation of the Gromov-Monge (GM) problem rooted in one of its fundamental properties: invariance to strong isomorphisms. We operationalize this property by decomposing the learnable OT map into two components: (i) an approximate strong isomorphism between the source distribution and an intermediate reference distribution, and (ii) a GM-optimal map between this reference and the target distribution. Our formulation leverages and extends the Monge gap regularizer of Uscidda & Cuturi (2023) to eliminate the need for complex architectural requirements of other neural OT methods, yielding a simple but practical method that enjoys favorable theoretical guarantees. Our preliminary empirical results show that our framework provides a promising approach to learn OT maps across diverse spaces.
Authors: Navid Seidi, Satyaki Roy, Sajal K. Das, Ardhendu Tripathy
Abstract: The diversity in disease profiles and therapeutic approaches between hospitals and health professionals underscores the need for patient-centric personalized strategies in healthcare. Alongside this, similarities in disease progression across patients can be utilized to improve prediction models in survival analysis. The need for patient privacy and the utility of prediction models can be simultaneously addressed in the framework of Federated Learning (FL). This paper outlines an approach in the domain of federated survival analysis, specifically the Cox Proportional Hazards (CoxPH) model, with a specific focus on mitigating data heterogeneity and elevating model performance. We present an FL approach that employs feature-based clustering to enhance model accuracy across synthetic datasets and real-world applications, including the Surveillance, Epidemiology, and End Results (SEER) database. Furthermore, we consider an event-based reporting strategy that provides a dynamic approach to model adaptation by responding to local data changes. Our experiments show the efficacy of our approach and discuss future directions for a practical application of FL in healthcare.
Authors: Vineet Thumuluri, Peter Eckmann, Michael K. Gilson, Rose Yu
Abstract: This technical report investigates variants of the Latent Inceptionism on Molecules (LIMO) framework to improve the properties of generated molecules. We conduct ablative studies of molecular representation, decoder model, and surrogate model training scheme. The experiments suggest that an autogressive Transformer decoder with GroupSELFIES achieves the best average properties for the random generation task.
Authors: Phuong Quynh Le, J\"org Schl\"otterer, Christin Seifert
Abstract: Machine learning models are known to learn spurious correlations, i.e., features having strong relations with class labels but no causal relation. Relying on those correlations leads to poor performance in the data groups without these correlations and poor generalization ability. To improve the robustness of machine learning models to spurious correlations, we propose an approach to extract a subnetwork from a fully trained network that does not rely on spurious correlations. The subnetwork is found by the assumption that data points with the same spurious attribute will be close to each other in the representation space when training with ERM, then we employ supervised contrastive loss in a novel way to force models to unlearn the spurious connections. The increase in the worst-group performance of our approach contributes to strengthening the hypothesis that there exists a subnetwork in a fully trained dense network that is responsible for using only invariant features in classification tasks, therefore erasing the influence of spurious features even in the setup of multi spurious attributes and no prior knowledge of attributes labels.
Authors: Vinod Kumar Maddineni, Naga Babu Koganti, Praveen Damacharla
Abstract: In this research, an effort is made to address microgrid systems' operational challenges, characterized by power oscillations that eventually contribute to grid instability. An integrated strategy is proposed, leveraging the strengths of convolutional and Gated Recurrent Unit (GRU) layers. This approach is aimed at effectively extracting temporal data from energy datasets to improve the precision of microgrid behavior forecasts. Additionally, an attention layer is employed to underscore significant features within the time-series data, optimizing the forecasting process. The framework is anchored by a Multi-Layer Perceptron (MLP) model, which is tasked with comprehensive load forecasting and the identification of abnormal grid behaviors. Our methodology underwent rigorous evaluation using the Micro-grid Tariff Assessment Tool dataset, with Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and the coefficient of determination (r2-score) serving as the primary metrics. The approach demonstrated exemplary performance, evidenced by a MAE of 0.39, RMSE of 0.28, and an r2-score of 98.89\% in load forecasting, along with near-perfect zero state prediction accuracy (approximately 99.9\%). Significantly outperforming conventional machine learning models such as support vector regression and random forest regression, our model's streamlined architecture is particularly suitable for real-time applications, thereby facilitating more effective and reliable microgrid management.
Authors: Ajay Jaiswal, Nurendra Choudhary, Ravinarayana Adkathimar, Muthu P. Alagappan, Gaurush Hiranandani, Ying Ding, Zhangyang Wang, Edward W Huang, Karthik Subbian
Abstract: Graph Neural Networks (GNNs) have attracted immense attention in the past decade due to their numerous real-world applications built around graph-structured data. On the other hand, Large Language Models (LLMs) with extensive pretrained knowledge and powerful semantic comprehension abilities have recently shown a remarkable ability to benefit applications using vision and text data. In this paper, we investigate how LLMs can be leveraged in a computationally efficient fashion to benefit rich graph-structured data, a modality relatively unexplored in LLM literature. Prior works in this area exploit LLMs to augment every node features in an ad-hoc fashion (not scalable for large graphs), use natural language to describe the complex structural information of graphs, or perform computationally expensive finetuning of LLMs in conjunction with GNNs. We propose E-LLaGNN (Efficient LLMs augmented GNNs), a framework with an on-demand LLM service that enriches message passing procedure of graph learning by enhancing a limited fraction of nodes from the graph. More specifically, E-LLaGNN relies on sampling high-quality neighborhoods using LLMs, followed by on-demand neighborhood feature enhancement using diverse prompts from our prompt catalog, and finally information aggregation using message passing from conventional GNN architectures. We explore several heuristics-based active node selection strategies to limit the computational and memory footprint of LLMs when handling millions of nodes. Through extensive experiments & ablation on popular graph benchmarks of varying scales (Cora, PubMed, ArXiv, & Products), we illustrate the effectiveness of our E-LLaGNN framework and reveal many interesting capabilities such as improved gradient flow in deep GNNs, LLM-free inference ability etc.
Authors: Dylan J. Foster, Adam Block, Dipendra Misra
Abstract: Imitation learning (IL) aims to mimic the behavior of an expert in a sequential decision making task by learning from demonstrations, and has been widely applied to robotics, autonomous driving, and autoregressive text generation. The simplest approach to IL, behavior cloning (BC), is thought to incur sample complexity with unfavorable quadratic dependence on the problem horizon, motivating a variety of different online algorithms that attain improved linear horizon dependence under stronger assumptions on the data and the learner's access to the expert. We revisit the apparent gap between offline and online IL from a learning-theoretic perspective, with a focus on general policy classes up to and including deep neural networks. Through a new analysis of behavior cloning with the logarithmic loss, we show that it is possible to achieve horizon-independent sample complexity in offline IL whenever (i) the range of the cumulative payoffs is controlled, and (ii) an appropriate notion of supervised learning complexity for the policy class is controlled. Specializing our results to deterministic, stationary policies, we show that the gap between offline and online IL is not fundamental: (i) it is possible to achieve linear dependence on horizon in offline IL under dense rewards (matching what was previously only known to be achievable in online IL); and (ii) without further assumptions on the policy class, online IL cannot improve over offline IL with the logarithmic loss, even in benign MDPs. We complement our theoretical results with experiments on standard RL tasks and autoregressive language generation to validate the practical relevance of our findings.
Authors: Beibei Li, Yiyuan Zheng, Beihong Jin, Tao Xiang, Haobo Wang, Lei Feng
Abstract: Partial-Label Learning (PLL) is a typical problem of weakly supervised learning, where each training instance is annotated with a set of candidate labels. Self-training PLL models achieve state-of-the-art performance but suffer from error accumulation problem caused by mistakenly disambiguated instances. Although co-training can alleviate this issue by training two networks simultaneously and allowing them to interact with each other, most existing co-training methods train two structurally identical networks with the same task, i.e., are symmetric, rendering it insufficient for them to correct each other due to their similar limitations. Therefore, in this paper, we propose an asymmetric dual-task co-training PLL model called AsyCo, which forces its two networks, i.e., a disambiguation network and an auxiliary network, to learn from different views explicitly by optimizing distinct tasks. Specifically, the disambiguation network is trained with self-training PLL task to learn label confidence, while the auxiliary network is trained in a supervised learning paradigm to learn from the noisy pairwise similarity labels that are constructed according to the learned label confidence. Finally, the error accumulation problem is mitigated via information distillation and confidence refinement. Extensive experiments on both uniform and instance-dependent partially labeled datasets demonstrate the effectiveness of AsyCo. The code is available at https://github.com/libeibeics/AsyCo.
Authors: Yi Liu, Chengjun Cai, Xiaoli Zhang, Xingliang Yuan, Cong Wang
Abstract: Large Vision Language Models (VLMs) extend and enhance the perceptual abilities of Large Language Models (LLMs). Despite offering new possibilities for LLM applications, these advancements raise significant security and ethical concerns, particularly regarding the generation of harmful content. While LLMs have undergone extensive security evaluations with the aid of red teaming frameworks, VLMs currently lack a well-developed one. To fill this gap, we introduce Arondight, a standardized red team framework tailored specifically for VLMs. Arondight is dedicated to resolving issues related to the absence of visual modality and inadequate diversity encountered when transitioning existing red teaming methodologies from LLMs to VLMs. Our framework features an automated multi-modal jailbreak attack, wherein visual jailbreak prompts are produced by a red team VLM, and textual prompts are generated by a red team LLM guided by a reinforcement learning agent. To enhance the comprehensiveness of VLM security evaluation, we integrate entropy bonuses and novelty reward metrics. These elements incentivize the RL agent to guide the red team LLM in creating a wider array of diverse and previously unseen test cases. Our evaluation of ten cutting-edge VLMs exposes significant security vulnerabilities, particularly in generating toxic images and aligning multi-modal prompts. In particular, our Arondight achieves an average attack success rate of 84.5\% on GPT-4 in all fourteen prohibited scenarios defined by OpenAI in terms of generating toxic text. For a clearer comparison, we also categorize existing VLMs based on their safety levels and provide corresponding reinforcement recommendations. Our multimodal prompt dataset and red team code will be released after ethics committee approval. CONTENT WARNING: THIS PAPER CONTAINS HARMFUL MODEL RESPONSES.
Authors: Logan Weber, Jesse Michel, Alex Renda, Michael Carbin
Abstract: A $\textit{neural surrogate of a program}$ is a neural network that mimics the behavior of a program. Researchers have used these neural surrogates to automatically tune program inputs, adapt programs to new settings, and accelerate computations. Researchers traditionally develop neural surrogates by training on input-output examples from a single program. Alternatively, language models trained on a large dataset including many programs can consume program text, to act as a neural surrogate. Using a language model to both generate a surrogate and act as a surrogate, however, leading to a trade-off between resource consumption and accuracy. We present $\textit{neural surrogate compilation}$, a technique for producing neural surrogates directly from program text without coupling neural surrogate generation and execution. We implement neural surrogate compilers using hypernetworks trained on a dataset of C programs and find that they produce neural surrogates that are $1.9$-$9.5\times$ as data-efficient, produce visual results that are $1.0$-$1.3\times$ more similar to ground truth, and train in $4.3$-$7.3\times$ fewer epochs than neural surrogates trained from scratch.
Authors: Yuxuan Jiang, Yujie Yang, Zhiqian Lan, Guojian Zhan, Shengbo Eben Li, Qi Sun, Jian Ma, Tianwen Yu, Changwu Zhang
Abstract: Rocket recycling is a crucial pursuit in aerospace technology, aimed at reducing costs and environmental impact in space exploration. The primary focus centers on rocket landing control, involving the guidance of a nonlinear underactuated rocket with limited fuel in real-time. This challenging task prompts the application of reinforcement learning (RL), yet goal-oriented nature of the problem poses difficulties for standard RL algorithms due to the absence of intermediate reward signals. This paper, for the first time, significantly elevates the success rate of rocket landing control from 8% with a baseline controller to 97% on a high-fidelity rocket model using RL. Our approach, called Random Annealing Jump Start (RAJS), is tailored for real-world goal-oriented problems by leveraging prior feedback controllers as guide policy to facilitate environmental exploration and policy learning in RL. In each episode, the guide policy navigates the environment for the guide horizon, followed by the exploration policy taking charge to complete remaining steps. This jump-start strategy prunes exploration space, rendering the problem more tractable to RL algorithms. The guide horizon is sampled from a uniform distribution, with its upper bound annealing to zero based on performance metrics, mitigating distribution shift and mismatch issues in existing methods. Additional enhancements, including cascading jump start, refined reward and terminal condition, and action smoothness regulation, further improve policy performance and practical applicability. The proposed method is validated through extensive evaluation and Hardware-in-the-Loop testing, affirming the effectiveness, real-time feasibility, and smoothness of the proposed controller.
Authors: Jiaxiang Yi, Ji Cheng, Miguel A. Bessa
Abstract: Multi-fidelity machine learning methods address the accuracy-efficiency trade-off by integrating scarce, resource-intensive high-fidelity data with abundant but less accurate low-fidelity data. We propose a practical multi-fidelity strategy for problems spanning low- and high-dimensional domains, integrating a non-probabilistic regression model for the low-fidelity with a Bayesian model for the high-fidelity. The models are trained in a staggered scheme, where the low-fidelity model is transfer-learned to the high-fidelity data and a Bayesian model is trained for the residual. This three-model strategy -- deterministic low-fidelity, transfer learning, and Bayesian residual -- leads to a prediction that includes uncertainty quantification both for noisy and noiseless multi-fidelity data. The strategy is general and unifies the topic, highlighting the expressivity trade-off between the transfer-learning and Bayesian models (a complex transfer-learning model leads to a simpler Bayesian model, and vice versa). We propose modeling choices for two scenarios, and argue in favor of using a linear transfer-learning model that fuses 1) kernel ridge regression for low-fidelity with Gaussian processes for high-fidelity; or 2) deep neural network for low-fidelity with a Bayesian neural network for high-fidelity. We demonstrate the effectiveness and efficiency of the proposed strategies and contrast them with the state-of-the-art based on various numerical examples. The simplicity of these formulations makes them practical for a broad scope of future engineering applications.
Authors: Giacomo Spigler
Abstract: We introduce Proximal Policy Distillation (PPD), a novel policy distillation method that integrates student-driven distillation and Proximal Policy Optimization (PPO) to increase sample efficiency and to leverage the additional rewards that the student policy collects during distillation. To assess the efficacy of our method, we compare PPD with two common alternatives, student-distill and teacher-distill, over a wide range of reinforcement learning environments that include discrete actions and continuous control (ATARI, Mujoco, and Procgen). For each environment and method, we perform distillation to a set of target student neural networks that are smaller, identical (self-distillation), or larger than the teacher network. Our findings indicate that PPD improves sample efficiency and produces better student policies compared to typical policy distillation approaches. Moreover, PPD demonstrates greater robustness than alternative methods when distilling policies from imperfect demonstrations. The code for the paper is released as part of a new Python library built on top of stable-baselines3 to facilitate policy distillation: `sb3-distill'.
Authors: Niels uit de Bos, Adri\`a Garriga-Alonso
Abstract: Circuits are supposed to accurately describe how a neural network performs a specific task, but do they really? We evaluate three circuits found in the literature (IOI, greater-than, and docstring) in an adversarial manner, considering inputs where the circuit's behavior maximally diverges from the full model. Concretely, we measure the KL divergence between the full model's output and the circuit's output, calculated through resample ablation, and we analyze the worst-performing inputs. Our results show that the circuits for the IOI and docstring tasks fail to behave similarly to the full model even on completely benign inputs from the original task, indicating that more robust circuits are needed for safety-critical applications.
Authors: Sanyam Vyas, Chris Hicks, Vasilios Mavroudis
Abstract: This paper investigates the threat of backdoors in Deep Reinforcement Learning (DRL) agent policies and proposes a novel method for their detection at runtime. Our study focuses on elusive in-distribution backdoor triggers. Such triggers are designed to induce a deviation in the behaviour of a backdoored agent while blending into the expected data distribution to evade detection. Through experiments conducted in the Atari Breakout environment, we demonstrate the limitations of current sanitisation methods when faced with such triggers and investigate why they present a challenging defence problem. We then evaluate the hypothesis that backdoor triggers might be easier to detect in the neural activation space of the DRL agent's policy network. Our statistical analysis shows that indeed the activation patterns in the agent's policy network are distinct in the presence of a trigger, regardless of how well the trigger is concealed in the environment. Based on this, we propose a new defence approach that uses a classifier trained on clean environment samples and detects abnormal activations. Our results show that even lightweight classifiers can effectively prevent malicious actions with considerable accuracy, indicating the potential of this research direction even against sophisticated adversaries.
Authors: Byeong Tak Lee, Joon-myoung Kwon, Yong-Yeon Jo
Abstract: Domain generalization involves training machine learning models to perform robustly on unseen samples from out-of-distribution datasets. Adversarial Data Augmentation (ADA) is a commonly used approach that enhances model adaptability by incorporating synthetic samples, designed to simulate potential unseen samples. While ADA effectively addresses amplitude-related distribution shifts, it falls short in managing temporal shifts, which are essential for time series data. To address this limitation, we propose the Temporal Adversarial Data Augmentation for time teries Data (TADA), which incorporates a time warping technique specifically targeting temporal shifts. Recognizing the challenge of non-differentiability in traditional time warping, we make it differentiable by leveraging phase shifts in the frequency domain. Our evaluations across diverse domains demonstrate that TADA significantly outperforms existing ADA variants, enhancing model performance across time series datasets with varied distributions.
Authors: Joshua Shay Kricheli, Khoa Vo, Aniruddha Datta, Spencer Ozgur, Paulo Shakarian
Abstract: Recent advances in Hierarchical Multi-label Classification (HMC), particularly neurosymbolic-based approaches, have demonstrated improved consistency and accuracy by enforcing constraints on a neural model during training. However, such work assumes the existence of such constraints a-priori. In this paper, we relax this strong assumption and present an approach based on Error Detection Rules (EDR) that allow for learning explainable rules about the failure modes of machine learning models. We show that these rules are not only effective in detecting when a machine learning classifier has made an error but also can be leveraged as constraints for HMC, thereby allowing the recovery of explainable constraints even if they are not provided. We show that our approach is effective in detecting machine learning errors and recovering constraints, is noise tolerant, and can function as a source of knowledge for neurosymbolic models on multiple datasets, including a newly introduced military vehicle recognition dataset.
Authors: Tae-Geun Kim
Abstract: This study proposes two novel learning rate schedulers: the Hyperbolic Learning Rate Scheduler (HyperbolicLR) and the Exponential Hyperbolic Learning Rate Scheduler (ExpHyperbolicLR). These schedulers attempt to address the inconsistent learning curves often observed in conventional schedulers when adjusting the number of epochs. By leveraging the asymptotic behavior of hyperbolic curves, the proposed schedulers maintain more consistent learning curves across varying epoch settings. The HyperbolicLR algorithm directly applies this property to the epoch-learning rate space, while the ExpHyperbolicLR maps this concept onto the exponential space of epochs and learning rates. To evaluate the performance of these schedulers, first we found the optimal hyperparameters for each scheduler on a small number of epochs, fixed these values, and compared their performance as the number of epochs increased. Our experimental results on various deep learning tasks and architectures demonstrate that both HyperbolicLR and ExpHyperbolicLR maintain more consistent performance improvements compared to conventional schedulers as the number of epochs increases. These findings suggest that our hyperbolic-based learning rate schedulers offer a more robust and efficient approach to training deep neural networks, especially in scenarios where computational resources or time constraints limit extensive hyperparameter searches.
Authors: Luca Corbucci, Mikko A Heikkila, David Solans Noguero, Anna Monreale, Nicolas Kourtellis
Abstract: Training and deploying Machine Learning models that simultaneously adhere to principles of fairness and privacy while ensuring good utility poses a significant challenge. The interplay between these three factors of trustworthiness is frequently underestimated and remains insufficiently explored. Consequently, many efforts focus on ensuring only two of these factors, neglecting one in the process. The decentralization of the datasets and the variations in distributions among the clients exacerbate the complexity of achieving this ethical trade-off in the context of Federated Learning (FL). For the first time in FL literature, we address these three factors of trustworthiness. We introduce PUFFLE, a high-level parameterised approach that can help in the exploration of the balance between utility, privacy, and fairness in FL scenarios. We prove that PUFFLE can be effective across diverse datasets, models, and data distributions, reducing the model unfairness up to 75%, with a maximum reduction in the utility of 17% in the worst-case scenario, while maintaining strict privacy guarantees during the FL training.
Authors: Hugo Inzirillo
Abstract: We explore various neural network architectures for modeling the dynamics of the cryptocurrency market. Traditional linear models often fall short in accurately capturing the unique and complex dynamics of this market. In contrast, Deep Neural Networks (DNNs) have demonstrated considerable proficiency in time series forecasting. This papers introduces novel neural network framework that blend the principles of econometric state space models with the dynamic capabilities of Recurrent Neural Networks (RNNs). We propose state space models using Long Short Term Memory (LSTM), Gated Residual Units (GRU) and Temporal Kolmogorov-Arnold Networks (TKANs). According to the results, TKANs, inspired by Kolmogorov-Arnold Networks (KANs) and LSTM, demonstrate promising outcomes.
Authors: Junn Yong Loo, Michelle Adeline, Arghya Pal, Vishnu Monn Baskaran, Chee-Ming Ting, Raphael C. -W. Phan
Abstract: Energy based models (EBMs) are appealing for their generality and simplicity in data likelihood modeling, but have conventionally been difficult to train due to the unstable and time-consuming implicit MCMC sampling during contrastive divergence training. In this paper, we present a novel energy-based generative framework, Variational Potential Flow (VAPO), that entirely dispenses with implicit MCMC sampling and does not rely on complementary latent models or cooperative training. The VAPO framework aims to learn a potential energy function whose gradient (flow) guides the prior samples, so that their density evolution closely follows an approximate data likelihood homotopy. An energy loss function is then formulated to minimize the Kullback-Leibler divergence between density evolution of the flow-driven prior and the data likelihood homotopy. Images can be generated after training the potential energy, by initializing the samples from Gaussian prior and solving the ODE governing the potential flow on a fixed time interval using generic ODE solvers. Experiment results show that the proposed VAPO framework is capable of generating realistic images on various image datasets. In particular, our proposed framework achieves competitive FID scores for unconditional image generation on the CIFAR-10 and CelebA datasets.
Authors: Ranga Shaarad Ayyagari, Anurita Ghosh, Ambedkar Dukkipati
Abstract: Standard reinforcement learning algorithms with a single policy perform poorly on tasks in complex environments involving sparse rewards, diverse behaviors, or long-term planning. This led to the study of algorithms that incorporate temporal abstraction by training a hierarchy of policies that plan over different time scales. The options framework has been introduced to implement such temporal abstraction by learning low-level options that act as extended actions controlled by a high-level policy. The main challenge in applying these algorithms to real-world problems is that they suffer from high sample complexity to train multiple levels of the hierarchy, which is impossible in online settings. Motivated by this, in this paper, we propose an offline hierarchical RL method that can learn options from existing offline datasets collected by other unknown agents. This is a very challenging problem due to the distribution mismatch between the learned options and the policies responsible for the offline dataset and to our knowledge, this is the first work in this direction. In this work, we propose a framework by which an online hierarchical reinforcement learning algorithm can be trained on an offline dataset of transitions collected by an unknown behavior policy. We validate our method on Gym MuJoCo locomotion environments and robotic gripper block-stacking tasks in the standard as well as transfer and goal-conditioned settings.
Authors: Yizi Zhang, Jingyan Shen, Xiaoxue Xiong, Yongchan Kwon
Abstract: Evaluating the contribution of individual data points to a model's prediction is critical for interpreting model predictions and improving model performance. Existing data contribution methods have been applied to various data types, including tabular data, images, and texts; however, their primary focus has been on i.i.d. settings. Despite the pressing need for principled approaches tailored to time series datasets, the problem of estimating data contribution in such settings remains unexplored, possibly due to challenges associated with handling inherent temporal dependencies. This paper introduces TimeInf, a data contribution estimation method for time-series datasets. TimeInf uses influence functions to attribute model predictions to individual time points while preserving temporal structures. Our extensive empirical results demonstrate that TimeInf outperforms state-of-the-art methods in identifying harmful anomalies and helpful time points for forecasting. Additionally, TimeInf offers intuitive and interpretable attributions of data values, allowing us to easily distinguish diverse anomaly patterns through visualizations.
Authors: Xuexin Chen, Ruichu Cai, Kaitao Zheng, Zhifan Jiang, Zhengting Huang, Zhifeng Hao, Zijian Li
Abstract: Graph Out-of-Distribution (OOD), requiring that models trained on biased data generalize to the unseen test data, has considerable real-world applications. One of the most mainstream methods is to extract the invariant subgraph by aligning the original and augmented data with the help of environment augmentation. However, these solutions might lead to the loss or redundancy of semantic subgraphs and result in suboptimal generalization. To address this challenge, we propose exploiting Probability of Necessity and Sufficiency (PNS) to extract sufficient and necessary invariant substructures. Beyond that, we further leverage the domain variant subgraphs related to the labels to boost the generalization performance in an ensemble manner. Specifically, we first consider the data generation process for graph data. Under mild conditions, we show that the sufficient and necessary invariant subgraph can be extracted by minimizing an upper bound, built on the theoretical advance of the probability of necessity and sufficiency. To further bridge the theory and algorithm, we devise the model called Sufficiency and Necessity Inspired Graph Learning (SNIGL), which ensembles an invariant subgraph classifier on top of latent sufficient and necessary invariant subgraphs, and a domain variant subgraph classifier specific to the test domain for generalization enhancement. Experimental results demonstrate that our SNIGL model outperforms the state-of-the-art techniques on six public benchmarks, highlighting its effectiveness in real-world scenarios.
Authors: Fr\'ed\'eric Zheng, Alexandre Proutiere
Abstract: We study the split Conformal Prediction method when applied to Markovian data. We quantify the gap in terms of coverage induced by the correlations in the data (compared to exchangeable data). This gap strongly depends on the mixing properties of the underlying Markov chain, and we prove that it typically scales as $\sqrt{t_\mathrm{mix}\ln(n)/n}$ (where $t_\mathrm{mix}$ is the mixing time of the chain). We also derive upper bounds on the impact of the correlations on the size of the prediction set. Finally we present $K$-split CP, a method that consists in thinning the calibration dataset and that adapts to the mixing properties of the chain. Its coverage gap is reduced to $t_\mathrm{mix}/(n\ln(n))$ without really affecting the size of the prediction set. We finally test our algorithms on synthetic and real-world datasets.
Authors: Sheila Schoepp, Mehran Taghian, Shotaro Miwa, Yoshihiro Mitsuka, Shadan Golestan, Osmar Za\"iane
Abstract: Industry is rapidly moving towards fully autonomous and interconnected systems that can detect and adapt to changing conditions, including machine hardware faults. Traditional methods for adding hardware fault tolerance to machines involve duplicating components and algorithmically reconfiguring a machine's processes when a fault occurs. However, the growing interest in reinforcement learning-based robotic control offers a new perspective on achieving hardware fault tolerance. However, limited research has explored the potential of these approaches for hardware fault tolerance in machines. This paper investigates the potential of two state-of-the-art reinforcement learning algorithms, Proximal Policy Optimization (PPO) and Soft Actor-Critic (SAC), to enhance hardware fault tolerance into machines. We assess the performance of these algorithms in two OpenAI Gym simulated environments, Ant-v2 and FetchReach-v1. Robot models in these environments are subjected to six simulated hardware faults. Additionally, we conduct an ablation study to determine the optimal method for transferring an agent's knowledge, acquired through learning in a normal (pre-fault) environment, to a (post-)fault environment in a continual learning setting. Our results demonstrate that reinforcement learning-based approaches can enhance hardware fault tolerance in simulated machines, with adaptation occurring within minutes. Specifically, PPO exhibits the fastest adaptation when retaining the knowledge within its models, while SAC performs best when discarding all acquired knowledge. Overall, this study highlights the potential of reinforcement learning-based approaches, such as PPO and SAC, for hardware fault tolerance in machines. These findings pave the way for the development of robust and adaptive machines capable of effectively operating in real-world scenarios.
Authors: Mounir Ghogho
Abstract: We delve into the issue of node classification within graphs, specifically reevaluating the concept of neighborhood aggregation, which is a fundamental component in graph neural networks (GNNs). Our analysis reveals conceptual flaws within certain benchmark GNN models when operating under the assumption of edge-independent node labels, a condition commonly observed in benchmark graphs employed for node classification. Approaching neighborhood aggregation from a statistical signal processing perspective, our investigation provides novel insights which may be used to design more efficient GNN models.
Authors: Parsa Razmara, Tina Khezresmaeilzadeh, B. Keith Jenkins
Abstract: The COVID-19 pandemic has underscored the necessity for advanced diagnostic tools in global health systems. Infrared Thermography (IRT) has proven to be a crucial non-contact method for measuring body temperature, vital for identifying febrile conditions associated with infectious diseases like COVID-19. Traditional non-contact infrared thermometers (NCITs) often exhibit significant variability in readings. To address this, we integrated machine learning algorithms with IRT to enhance the accuracy and reliability of temperature measurements. Our study systematically evaluated various regression models using heuristic feature engineering techniques, focusing on features' physiological relevance and statistical significance. The Convolutional Neural Network (CNN) model, utilizing these techniques, achieved the lowest RMSE of 0.2223, demonstrating superior performance compared to results reported in previous literature. Among non-neural network models, the Binning method achieved the best performance with an RMSE of 0.2296. Our findings highlight the potential of combining advanced feature engineering with machine learning to improve diagnostic tools' effectiveness, with implications extending to other non-contact or remote sensing biomedical applications. This paper offers a comprehensive analysis of these methodologies, providing a foundation for future research in the field of non-invasive medical diagnostics.
Authors: Jiaxing Zhang, Jiayi Liu, Dongsheng Luo, Jennifer Neville, Hua Wei
Abstract: Recent studies seek to provide Graph Neural Network (GNN) interpretability via multiple unsupervised learning models. Due to the scarcity of datasets, current methods easily suffer from learning bias. To solve this problem, we embed a Large Language Model (LLM) as knowledge into the GNN explanation network to avoid the learning bias problem. We inject LLM as a Bayesian Inference (BI) module to mitigate learning bias. The efficacy of the BI module has been proven both theoretically and experimentally. We conduct experiments on both synthetic and real-world datasets. The innovation of our work lies in two parts: 1. We provide a novel view of the possibility of an LLM functioning as a Bayesian inference to improve the performance of existing algorithms; 2. We are the first to discuss the learning bias issues in the GNN explanation problem.
Authors: Hanxi Guo, Hao Wang, Tao Song, Tianhang Zheng, Yang Hua, Haibing Guan, Xiangyu Zhang
Abstract: Without direct access to the client's data, federated learning (FL) is well-known for its unique strength in data privacy protection among existing distributed machine learning techniques. However, its distributive and iterative nature makes FL inherently vulnerable to various poisoning attacks. To counteract these threats, extensive defenses have been proposed to filter out malicious clients, using various detection metrics. Based on our analysis of existing attacks and defenses, we find that there is a lack of attention to model redundancy. In neural networks, various model parameters contribute differently to the model's performance. However, existing attacks in FL manipulate all the model update parameters with the same strategy, making them easily detectable by common defenses. Meanwhile, the defenses also tend to analyze the overall statistical features of the entire model updates, leaving room for sophisticated attacks. Based on these observations, this paper proposes a generic and attack-agnostic augmentation approach designed to enhance the effectiveness and stealthiness of existing FL poisoning attacks against detection in FL, pointing out the inherent flaws of existing defenses and exposing the necessity of fine-grained FL security. Specifically, we employ a three-stage methodology that strategically constructs, generates, and injects poison (generated by existing attacks) into a pill (a tiny subnet with a novel structure) during the FL training, named as pill construction, pill poisoning, and pill injection accordingly. Extensive experimental results show that FL poisoning attacks enhanced by our method can bypass all the popular defenses, and can gain an up to 7x error rate increase, as well as on average a more than 2x error rate increase on both IID and non-IID data, in both cross-silo and cross-device FL systems.
Authors: Andrea Augello, Ashish Gupta, Giuseppe Lo Re, Sajal K. Das
Abstract: Federated Learning (FL) is a distributed machine learning paradigm facilitating participants to collaboratively train a model without revealing their local data. However, when FL is deployed into the wild, some intelligent clients can deliberately deviate from the standard training process to make the global model inclined toward their local model, thereby prioritizing their local data distribution. We refer to this novel category of misbehaving clients as selfish. In this paper, we propose a Robust aggregation strategy for FL server to mitigate the effect of Selfishness (in short RFL-Self). RFL-Self incorporates an innovative method to recover (or estimate) the true updates of selfish clients from the received ones, leveraging robust statistics (median of norms) of the updates at every round. By including the recovered updates in aggregation, our strategy offers strong robustness against selfishness. Our experimental results, obtained on MNIST and CIFAR-10 datasets, demonstrate that just 2% of clients behaving selfishly can decrease the accuracy by up to 36%, and RFL-Self can mitigate that effect without degrading the global model performance.
Authors: Jungang Yang, Zhe Ji, Liyao Xiang
Abstract: Differential Privacy (DP) mechanisms, especially in high-dimensional settings, often face the challenge of maintaining privacy without compromising the data utility. This work introduces an innovative shuffling mechanism in Differentially-Private Stochastic Gradient Descent (DPSGD) to enhance the utility of large models at the same privacy guarantee of the unshuffled case. Specifically, we reveal that random shuffling brings additional randomness to the trajectory of gradient descent while not impacting the model accuracy by the permutation invariance property -- the model can be equivalently computed in both forward and backward propagations under permutation. We show that permutation indeed improves the privacy guarantee of DPSGD in theory, but tracking the exact privacy loss on shuffled model is particularly challenging. Hence we exploit the approximation on sum of lognormal distributions to derive the condition for the shuffled DPSGD to meet the DP guarantee. Auditing results show that our condition offers a DP guarantee quite close to the audited privacy level, demonstrating our approach an effective estimation in practice. Experimental results have verified our theoretical derivation and illustrate that our mechanism improves the accuracy of DPSGD over the state-of-the-art baselines on a variety of models and tasks.
Authors: Adri\`a Garriga-Alonso, Mohammad Taufeeque, Adam Gleave
Abstract: To predict how advanced neural networks generalize to novel situations, it is essential to understand how they reason. Guez et al. (2019, "An investigation of model-free planning") trained a recurrent neural network (RNN) to play Sokoban with model-free reinforcement learning. They found that adding extra computation steps to the start of episodes at test time improves the RNN's success rate. We further investigate this phenomenon, finding that it rapidly emerges early on in training and then slowly fades, but only for comparatively easier levels. The RNN also often takes redundant actions at episode starts, and these are reduced by adding extra computation steps. Our results suggest that the RNN learns to take time to think by `pacing', despite the per-step penalties, indicating that training incentivizes planning capabilities. The small size (1.29M parameters) and interesting behavior of this model make it an excellent model organism for mechanistic interpretability.
Authors: Aki H\"arm\"a, Marcin Pietrasik, Anna Wilbik
Abstract: Large pretrained self-attention neural networks, or transformers, have been very successful in various tasks recently. The performance of a model on a given task depends on its ability to memorize and generalize the training data. Large transformer models, which may have billions of parameters, in theory have a huge capacity to memorize content. However, the current algorithms for the optimization fall short of the theoretical capacity, and the capacity is also highly dependent on the content. In this paper, we focus on the memory capacity of these models obtained using common training algorithms and synthetic training data. Based on the results, we derive an empirical capacity model (ECM) for a generic transformer. The ECM can be used to design task-specific transformer models with an optimal number of parameters in cases where the target memorization capability of the task can be defined.
Authors: Ye Lin Tun, Chu Myaet Thwal, Minh N. H. Nguyen, Choong Seon Hong
Abstract: Combining different data modalities enables deep neural networks to tackle complex tasks more effectively, making multimodal learning increasingly popular. To harness multimodal data closer to end users, it is essential to integrate multimodal learning with privacy-preserving training approaches such as federated learning (FL). However, compared to conventional unimodal learning, multimodal setting requires dedicated encoders for each modality, resulting in larger and more complex models that demand significant resources. This presents a substantial challenge for FL clients operating with limited computational resources and communication bandwidth. To address these challenges, we introduce LW-FedMML, a layer-wise federated multimodal learning approach, which decomposes the training process into multiple steps. Each step focuses on training only a portion of the model, thereby significantly reducing the memory and computational requirements. Moreover, FL clients only need to exchange the trained model portion with the central server, lowering the resulting communication cost. We conduct extensive experiments across various FL scenarios and multimodal learning setups to validate the effectiveness of our proposed method. The results demonstrate that LW-FedMML can compete with conventional end-to-end federated multimodal learning (FedMML) while significantly reducing the resource burden on FL clients. Specifically, LW-FedMML reduces memory usage by up to $2.7\times$, computational operations (FLOPs) by $2.4\times$, and total communication cost by $2.3\times$. We also introduce a progressive training approach called Prog-FedMML. While it offers lesser resource efficiency than LW-FedMML, Prog-FedMML has the potential to surpass the performance of end-to-end FedMML, making it a viable option for scenarios with fewer resource constraints.
Authors: Ziqun Chen, Kechao Cai, Zhuoyue Chen, Jinbei Zhang, John C. S. Lui
Abstract: We study the stochastic combinatorial semi-bandit problem with unrestricted feedback delays under merit-based fairness constraints. This is motivated by applications such as crowdsourcing, and online advertising, where immediate feedback is not immediately available and fairness among different choices (or arms) is crucial. We consider two types of unrestricted feedback delays: reward-independent delays where the feedback delays are independent of the rewards, and reward-dependent delays where the feedback delays are correlated with the rewards. Furthermore, we introduce merit-based fairness constraints to ensure a fair selection of the arms. We define the reward regret and the fairness regret and present new bandit algorithms to select arms under unrestricted feedback delays based on their merits. We prove that our algorithms all achieve sublinear expected reward regret and expected fairness regret, with a dependence on the quantiles of the delay distribution. We also conduct extensive experiments using synthetic and real-world data and show that our algorithms can fairly select arms with different feedback delays.
Authors: Vipul Gupta, Xin Chen, Ruoyun Huang, Fanlong Meng, Jianjun Chen, Yujun Yan
Abstract: Graph Neural Networks (GNNs) have emerged as powerful tools for supervised machine learning over graph-structured data, while sampling-based node representation learning is widely utilized in unsupervised learning. However, scalability remains a major challenge in both supervised and unsupervised learning for large graphs (e.g., those with over 1 billion nodes). The scalability bottleneck largely stems from the mini-batch sampling phase in GNNs and the random walk sampling phase in unsupervised methods. These processes often require storing features or embeddings in memory. In the context of distributed training, they require frequent, inefficient random access to data stored across different workers. Such repeated inter-worker communication for each mini-batch leads to high communication overhead and computational inefficiency. We propose GraphScale, a unified framework for both supervised and unsupervised learning to store and process large graph data distributedly. The key insight in our design is the separation of workers who store data and those who perform the training. This separation allows us to decouple computing and storage in graph training, thus effectively building a pipeline where data fetching and data computation can overlap asynchronously. Our experiments show that GraphScale outperforms state-of-the-art methods for distributed training of both GNNs and node embeddings. We evaluate GraphScale both on public and proprietary graph datasets and observe a reduction of at least 40% in end-to-end training times compared to popular distributed frameworks, without any loss in performance. While most existing methods don't support billion-node graphs for training node embeddings, GraphScale is currently deployed in production at TikTok enabling efficient learning over such large graphs.
Authors: Xinghao Wu, Xuefeng Liu, Jianwei Niu, Guogang Zhu, Shaojie Tang, Xiaotian Li, Jiannong Cao
Abstract: Personalized Federated Learning (PFL) is a commonly used framework that allows clients to collaboratively train their personalized models. PFL is particularly useful for handling situations where data from different clients are not independent and identically distributed (non-IID). Previous research in PFL implicitly assumes that clients can gain more benefits from those with similar data distributions. Correspondingly, methods such as personalized weight aggregation are developed to assign higher weights to similar clients during training. We pose a question: can a client benefit from other clients with dissimilar data distributions and if so, how? This question is particularly relevant in scenarios with a high degree of non-IID, where clients have widely different data distributions, and learning from only similar clients will lose knowledge from many other clients. We note that when dealing with clients with similar data distributions, methods such as personalized weight aggregation tend to enforce their models to be close in the parameter space. It is reasonable to conjecture that a client can benefit from dissimilar clients if we allow their models to depart from each other. Based on this idea, we propose DiversiFed which allows each client to learn from clients with diversified data distribution in personalized federated learning. DiversiFed pushes personalized models of clients with dissimilar data distributions apart in the parameter space while pulling together those with similar distributions. In addition, to achieve the above effect without using prior knowledge of data distribution, we design a loss function that leverages the model similarity to determine the degree of attraction and repulsion between any two models. Experiments on several datasets show that DiversiFed can benefit from dissimilar clients and thus outperform the state-of-the-art methods.
Authors: Peng Cheng, Huimu Wang, Jinyuan Zhao, Yihao Wang, Enqiang Xu, Yu Zhao, Zhuojian Xiao, Songlin Wang, Guoyu Tang, Lin Liu, Sulong Xu
Abstract: Traffic allocation is a process of redistributing natural traffic to products by adjusting their positions in the post-search phase, aimed at effectively fostering merchant growth, precisely meeting customer demands, and ensuring the maximization of interests across various parties within e-commerce platforms. Existing methods based on learning to rank neglect the long-term value of traffic allocation, whereas approaches of reinforcement learning suffer from balancing multiple objectives and the difficulties of cold starts within realworld data environments. To address the aforementioned issues, this paper propose a multi-objective deep reinforcement learning framework consisting of multi-objective Q-learning (MOQ), a decision fusion algorithm (DFM) based on the cross-entropy method(CEM), and a progressive data augmentation system(PDA). Specifically. MOQ constructs ensemble RL models, each dedicated to an objective, such as click-through rate, conversion rate, etc. These models individually determine the position of items as actions, aiming to estimate the long-term value of multiple objectives from an individual perspective. Then we employ DFM to dynamically adjust weights among objectives to maximize long-term value, addressing temporal dynamics in objective preferences in e-commerce scenarios. Initially, PDA trained MOQ with simulated data from offline logs. As experiments progressed, it strategically integrated real user interaction data, ultimately replacing the simulated dataset to alleviate distributional shifts and the cold start problem. Experimental results on real-world online e-commerce systems demonstrate the significant improvements of MODRL-TA, and we have successfully deployed MODRL-TA on an e-commerce search platform.
Authors: Adway Girish, Alliot Nagle, Marco Bondaschi, Michael Gastpar, Ashok Vardhan Makkuva, Hyeji Kim
Abstract: We formalize the problem of prompt compression for large language models (LLMs) and present a framework to unify token-level prompt compression methods which create hard prompts for black-box models. We derive the distortion-rate function for this setup as a linear program, and provide an efficient algorithm to compute this fundamental limit via the dual of the linear program. Using the distortion-rate function as the baseline, we study the performance of existing compression schemes on a synthetic dataset consisting of prompts generated from a Markov chain, natural language queries, and their respective answers. Our empirical analysis demonstrates the criticality of query-aware prompt compression, where the compressor has knowledge of the downstream task/query for the black-box LLM. We show that there is a large gap between the performance of current prompt compression methods and the optimal strategy, and propose a query-aware, variable-rate adaptation of a prior work to close the gap. We extend our experiments to a small natural language dataset to further confirm our findings on our synthetic dataset.
Authors: Francisco Mena, Diego Arenas, Andreas Dengel
Abstract: Multi-sensor ML models for EO aim to enhance prediction accuracy by integrating data from various sources. However, the presence of missing data poses a significant challenge, particularly in non-persistent sensors that can be affected by external factors. Existing literature has explored strategies like temporal dropout and sensor-invariant models to address the generalization to missing data issues. Inspired by these works, we study two novel methods tailored for multi-sensor scenarios, namely Input Sensor Dropout (ISensD) and Ensemble Sensor Invariant (ESensI). Through experimentation on three multi-sensor temporal EO datasets, we demonstrate that these methods effectively increase the robustness of model predictions to missing sensors. Particularly, we focus on how the predictive performance of models drops when sensors are missing at different levels. We observe that ensemble multi-sensor models are the most robust to the lack of sensors. In addition, the sensor dropout component in ISensD shows promising robustness results.
Authors: Georgy Tyukin, Gbetondji J-S Dovonon, Jean Kaddour, Pasquale Minervini
Abstract: The inference demand for LLMs has skyrocketed in recent months, and serving models with low latencies remains challenging due to the quadratic input length complexity of the attention layers. In this work, we investigate the effect of dropping MLP and attention layers at inference time on the performance of Llama-v2 models. We find that dropping dreeper attention layers only marginally decreases performance but leads to the best speedups alongside dropping entire layers. For example, removing 33\% of attention layers in a 13B Llama2 model results in a 1.8\% drop in average performance over the OpenLLM benchmark. We also observe that skipping layers except the latter layers reduces performances for more layers skipped, except for skipping the attention layers.
Authors: Corentin Sala\"un, Xingchang Huang, Iliyan Georgiev, Niloy J. Mitra, Gurprit Singh
Abstract: We introduce a theoretical and practical framework for efficient importance sampling of mini-batch samples for gradient estimation from single and multiple probability distributions. To handle noisy gradients, our framework dynamically evolves the importance distribution during training by utilizing a self-adaptive metric. Our framework combines multiple, diverse sampling distributions, each tailored to specific parameter gradients. This approach facilitates the importance sampling of vector-valued gradient estimation. Rather than naively combining multiple distributions, our framework involves optimally weighting data contribution across multiple distributions. This adapted combination of multiple importance yields superior gradient estimates, leading to faster training convergence. We demonstrate the effectiveness of our approach through empirical evaluations across a range of optimization tasks like classification and regression on both image and point cloud datasets.
Authors: Shiqing Gao, Jiaxin Ding, Luoyi Fu, Xinbing Wang, Chenghu Zhou
Abstract: In Constrained Reinforcement Learning (CRL), agents explore the environment to learn the optimal policy while satisfying constraints. The penalty function method has recently been studied as an effective approach for handling constraints, which imposes constraints penalties on the objective to transform the constrained problem into an unconstrained one. However, it is challenging to choose appropriate penalties that balance policy performance and constraint satisfaction efficiently. In this paper, we propose a theoretically guaranteed penalty function method, Exterior Penalty Policy Optimization (EPO), with adaptive penalties generated by a Penalty Metric Network (PMN). PMN responds appropriately to varying degrees of constraint violations, enabling efficient constraint satisfaction and safe exploration. We theoretically prove that EPO consistently improves constraint satisfaction with a convergence guarantee. We propose a new surrogate function and provide worst-case constraint violation and approximation error. In practice, we propose an effective smooth penalty function, which can be easily implemented with a first-order optimizer. Extensive experiments are conducted, showing that EPO outperforms the baselines in terms of policy performance and constraint satisfaction with a stable training process, particularly on complex tasks.
Authors: Georgii Novikov, Ivan Oseledets
Abstract: The scaling of neural networks with increasing data and model sizes necessitates more efficient deep learning algorithms. This paper addresses the memory footprint challenge in neural network training by proposing a modification to the handling of activation tensors in pointwise nonlinearity layers. Traditionally, these layers save the entire input tensor for the backward pass, leading to substantial memory use. Our method involves saving the output tensor instead, reducing the memory required when the subsequent layer also saves its input tensor. This approach is particularly beneficial for transformer-based architectures like GPT, BERT, Mistral, and Llama. Application of our method involves taken an inverse function of nonlinearity. To the best of our knowledge, that can not be done analitically and instead we buid an accurate approximations using simpler functions. Experimental results confirm that our method significantly reduces memory usage without affecting training accuracy. The implementation is available at https://github.com/PgLoLo/optiacts.
Authors: Abhay Sheshadri, Aidan Ewart, Phillip Guo, Aengus Lynch, Cindy Wu, Vivek Hebbar, Henry Sleight, Asa Cooper Stickland, Ethan Perez, Dylan Hadfield-Menell, Stephen Casper
Abstract: Large language models (LLMs) can often be made to behave in undesirable ways that they are explicitly fine-tuned not to. For example, the LLM red-teaming literature has produced a wide variety of `jailbreaking' techniques to elicit harmful text from models that were fine-tuned to be harmless. Recent work on red-teaming, model editing, and interpretability suggests that this challenge stems from how (adversarial) fine-tuning largely serves to suppress rather than remove undesirable capabilities from LLMs. Prior work has introduced latent adversarial training (LAT) as a way to improve robustness to broad classes of failures. These prior works have considered untargeted latent space attacks where the adversary perturbs latent activations to maximize loss on examples of desirable behavior. Untargeted LAT can provide a generic type of robustness but does not leverage information about specific failure modes. Here, we experiment with targeted LAT where the adversary seeks to minimize loss on a specific competing task. We find that it can augment a wide variety of state-of-the-art methods. First, we use targeted LAT to improve robustness to jailbreaks, outperforming a strong R2D2 baseline with orders of magnitude less compute. Second, we use it to more effectively remove backdoors with no knowledge of the trigger. Finally, we use it to more effectively unlearn knowledge for specific undesirable tasks in a way that is also more robust to re-learning. Overall, our results suggest that targeted LAT can be an effective tool for defending against harmful behaviors from LLMs.
Authors: Jiayi Wang, Shiqiang Wang, Rong-Rong Chen, Mingyue Ji
Abstract: In federated learning (FL), data heterogeneity is the main reason that existing theoretical analyses are pessimistic about the convergence rate. In particular, for many FL algorithms, the convergence rate grows dramatically when the number of local updates becomes large, especially when the product of the gradient divergence and local Lipschitz constant is large. However, empirical studies can show that more local updates can improve the convergence rate even when these two parameters are large, which is inconsistent with the theoretical findings. This paper aims to bridge this gap between theoretical understanding and practical performance by providing a theoretical analysis from a new perspective on data heterogeneity. In particular, we propose a new and weaker assumption compared to the local Lipschitz gradient assumption, named the heterogeneity-driven pseudo-Lipschitz assumption. We show that this and the gradient divergence assumptions can jointly characterize the effect of data heterogeneity. By deriving a convergence upper bound for FedAvg and its extensions, we show that, compared to the existing works, local Lipschitz constant is replaced by the much smaller heterogeneity-driven pseudo-Lipschitz constant and the corresponding convergence upper bound can be significantly reduced for the same number of local updates, although its order stays the same. In addition, when the local objective function is quadratic, more insights on the impact of data heterogeneity can be obtained using the heterogeneity-driven pseudo-Lipschitz constant. For example, we can identify a region where FedAvg can outperform mini-batch SGD even when the gradient divergence can be arbitrarily large. Our findings are validated using experiments.
Authors: David Perera, Victor Letzelter, Th\'eo Mariotte, Adrien Cort\'es, Mickael Chen, Slim Essid, Ga\"el Richard
Abstract: We introduce Annealed Multiple Choice Learning (aMCL) which combines simulated annealing with MCL. MCL is a learning framework handling ambiguous tasks by predicting a small set of plausible hypotheses. These hypotheses are trained using the Winner-takes-all (WTA) scheme, which promotes the diversity of the predictions. However, this scheme may converge toward an arbitrarily suboptimal local minimum, due to the greedy nature of WTA. We overcome this limitation using annealing, which enhances the exploration of the hypothesis space during training. We leverage insights from statistical physics and information theory to provide a detailed description of the model training trajectory. Additionally, we validate our algorithm by extensive experiments on synthetic datasets, on the standard UCI benchmark, and on speech separation.
Authors: Itai Gat, Tal Remez, Neta Shaul, Felix Kreuk, Ricky T. Q. Chen, Gabriel Synnaeve, Yossi Adi, Yaron Lipman
Abstract: Despite Flow Matching and diffusion models having emerged as powerful generative paradigms for continuous variables such as images and videos, their application to high-dimensional discrete data, such as language, is still limited. In this work, we present Discrete Flow Matching, a novel discrete flow paradigm designed specifically for generating discrete data. Discrete Flow Matching offers several key contributions: (i) it works with a general family of probability paths interpolating between source and target distributions; (ii) it allows for a generic formula for sampling from these probability paths using learned posteriors such as the probability denoiser ($x$-prediction) and noise-prediction ($\epsilon$-prediction); (iii) practically, focusing on specific probability paths defined with different schedulers considerably improves generative perplexity compared to previous discrete diffusion and flow models; and (iv) by scaling Discrete Flow Matching models up to 1.7B parameters, we reach 6.7% Pass@1 and 13.4% Pass@10 on HumanEval and 6.7% Pass@1 and 20.6% Pass@10 on 1-shot MBPP coding benchmarks. Our approach is capable of generating high-quality discrete data in a non-autoregressive fashion, significantly closing the gap between autoregressive models and discrete flow models.
Authors: Hossein Nematzadeh, Joseph Mani, Zahra Nematzadeh, Ebrahim Akbari, Radziah Mohamad
Abstract: Feature selection poses a challenge in small-sample high-dimensional datasets, where the number of features exceeds the number of observations, as seen in microarray, gene expression, and medical datasets. There isn't a universally optimal feature selection method applicable to any data distribution, and as a result, the literature consistently endeavors to address this issue. One recent approach in feature selection is termed frequency-based feature selection. However, existing methods in this domain tend to overlook feature values, focusing solely on the distribution in the response variable. In response, this paper introduces the Distance-based Mutual Congestion (DMC) as a filter method that considers both the feature values and the distribution of observations in the response variable. DMC sorts the features of datasets, and the top 5% are retained and clustered by KMeans to mitigate multicollinearity. This is achieved by randomly selecting one feature from each cluster. The selected features form the feature space, and the search space for the Genetic Algorithm with Adaptive Rates (GAwAR) will be approximated using this feature space. GAwAR approximates the combination of the top 10 features that maximizes prediction accuracy within a wrapper scheme. To prevent premature convergence, GAwAR adaptively updates the crossover and mutation rates. The hybrid DMC-GAwAR is applicable to binary classification datasets, and experimental results demonstrate its superiority over some recent works. The implementation and corresponding data are available at https://github.com/hnematzadeh/DMC-GAwAR
Authors: Henry W. Leung, Jo Bovy, Joshua S. Speagle
Abstract: Transformers are often the go-to architecture to build foundation models that ingest a large amount of training data. But these models do not estimate the probability density distribution when trained on regression problems, yet obtaining full probabilistic outputs is crucial to many fields of science, where the probability distribution of the answer can be non-Gaussian and multimodal. In this work, we demonstrate that training a probabilistic model using a denoising diffusion head on top of the Transformer provides reasonable probability density estimation even for high-dimensional inputs. The combined Transformer+Denoising Diffusion model allows conditioning the output probability density on arbitrary combinations of inputs and it is thus a highly flexible density function emulator of all possible input/output combinations. We illustrate our Transformer+Denoising Diffusion model by training it on a large dataset of astronomical observations and measured labels of stars within our Galaxy and we apply it to a variety of inference tasks to show that the model can infer labels accurately with reasonable distributions.
Authors: Mohammad Kohankhaki, Ahmad Ayad, Mahdi Barhoush, Anke Schmeink
Abstract: The expansion of IoT devices and the demands of Deep Learning have highlighted significant challenges in Distributed Deep Learning (DDL) systems. Parallel Split Learning (PSL) has emerged as a promising derivative of Split Learning that is well suited for distributed learning on resource-constrained devices. However, PSL faces several obstacles, such as large effective batch sizes, non-IID data distributions, and the straggler effect. We view these issues as a sampling dilemma and propose to address them by orchestrating the mini-batch sampling process on the server side. We introduce the Uniform Global Sampling (UGS) method to decouple the effective batch size from the number of clients and reduce mini-batch deviation in non-IID settings. To address the straggler effect, we introduce the Latent Dirichlet Sampling (LDS) method, which generalizes UGS to balance the trade-off between batch deviation and training time. Our simulations reveal that our proposed methods enhance model accuracy by up to 34.1% in non-IID settings and reduce the training time in the presence of stragglers by up to 62%. In particular, LDS effectively mitigates the straggler effect without compromising model accuracy or adding significant computational overhead compared to UGS. Our results demonstrate the potential of our methods as a promising solution for DDL in real applications.
Authors: Davis Brown, Cody Nizinski, Madelyn Shapiro, Corey Fallon, Tianzhixi Yin, Henry Kvinge, Jonathan H. Tu
Abstract: Deep learning still struggles with certain kinds of scientific data. Notably, pretraining data may not provide coverage of relevant distribution shifts (e.g., shifts induced via the use of different measurement instruments). We consider deep learning models trained to classify the synthesis conditions of uranium ore concentrates (UOCs) and show that model editing is particularly effective for improving generalization to distribution shifts common in this domain. In particular, model editing outperforms finetuning on two curated datasets comprising of micrographs taken of U$_{3}$O$_{8}$ aged in humidity chambers and micrographs acquired with different scanning electron microscopes, respectively.
Authors: Kaiwen Wang, Rahul Kidambi, Ryan Sullivan, Alekh Agarwal, Christoph Dann, Andrea Michi, Marco Gelmi, Yunxuan Li, Raghav Gupta, Avinava Dubey, Alexandre Ram\'e, Johan Ferret, Geoffrey Cideron, Le Hou, Hongkun Yu, Amr Ahmed, Aranyak Mehta, L\'eonard Hussenot, Olivier Bachem, Edouard Leurent
Abstract: Reward-based finetuning is crucial for aligning language policies with intended behaviors (e.g., creativity and safety). A key challenge here is to develop steerable language models that trade-off multiple (conflicting) objectives in a flexible and efficient manner. This paper presents Conditioned Language Policy (CLP), a general framework for finetuning language models on multiple objectives. Building on techniques from multi-task training and parameter-efficient finetuning, CLP can learn steerable models that effectively trade-off conflicting objectives at inference time. Notably, this does not require training or maintaining multiple models to achieve different trade-offs between the objectives. Through an extensive set of experiments and ablations, we show that the CLP framework learns steerable models that outperform and Pareto-dominate the current state-of-the-art approaches for multi-objective finetuning.
Authors: Yongcan Yu, Lijun Sheng, Ran He, Jian Liang
Abstract: Test-time adaptation (TTA) aims to address the distribution shift between the training and test data with only unlabeled data at test time. Existing TTA methods often focus on improving recognition performance specifically for test data associated with classes in the training set. However, during the open-world inference process, there are inevitably test data instances from unknown classes, commonly referred to as outliers. This paper pays attention to the problem that conducts both sample recognition and outlier rejection during inference while outliers exist. To address this problem, we propose a new approach called STAble Memory rePlay (STAMP), which performs optimization over a stable memory bank instead of the risky mini-batch. In particular, the memory bank is dynamically updated by selecting low-entropy and label-consistent samples in a class-balanced manner. In addition, we develop a self-weighted entropy minimization strategy that assigns higher weight to low-entropy samples. Extensive results demonstrate that STAMP outperforms existing TTA methods in terms of both recognition and outlier detection performance. The code is released at https://github.com/yuyongcan/STAMP.
Authors: Zhuorui Ye, Stephanie Milani, Geoffrey J. Gordon, Fei Fang
Abstract: Recent advances in reinforcement learning (RL) have predominantly leveraged neural network-based policies for decision-making, yet these models often lack interpretability, posing challenges for stakeholder comprehension and trust. Concept bottleneck models offer an interpretable alternative by integrating human-understandable concepts into neural networks. However, a significant limitation in prior work is the assumption that human annotations for these concepts are readily available during training, necessitating continuous real-time input from human annotators. To overcome this limitation, we introduce a novel training scheme that enables RL algorithms to efficiently learn a concept-based policy by only querying humans to label a small set of data, or in the extreme case, without any human labels. Our algorithm, LICORICE, involves three main contributions: interleaving concept learning and RL training, using a concept ensembles to actively select informative data points for labeling, and decorrelating the concept data with a simple strategy. We show how LICORICE reduces manual labeling efforts to to 500 or fewer concept labels in three environments. Finally, we present an initial study to explore how we can use powerful vision-language models to infer concepts from raw visual inputs without explicit labels at minimal cost to performance.
Authors: Daniil Dmitriev, Rares-Darius Buhai, Stefan Tiegel, Alexander Wolters, Gleb Novikov, Amartya Sanyal, David Steurer, Fanny Yang
Abstract: We study the problem of estimating the means of well-separated mixtures when an adversary may add arbitrary outliers. While strong guarantees are available when the outlier fraction is significantly smaller than the minimum mixing weight, much less is known when outliers may crowd out low-weight clusters - a setting we refer to as list-decodable mixture learning (LD-ML). In this case, adversarial outliers can simulate additional spurious mixture components. Hence, if all means of the mixture must be recovered up to a small error in the output list, the list size needs to be larger than the number of (true) components. We propose an algorithm that obtains order-optimal error guarantees for each mixture mean with a minimal list-size overhead, significantly improving upon list-decodable mean estimation, the only existing method that is applicable for LD-ML. Although improvements are observed even when the mixture is non-separated, our algorithm achieves particularly strong guarantees when the mixture is separated: it can leverage the mixture structure to partially cluster the samples before carefully iterating a base learner for list-decodable mean estimation at different scales.
Authors: Yakir Oz, Gilad Yehudai, Gal Vardi, Itai Antebi, Michal Irani, Niv Haim
Abstract: Current methods for reconstructing training data from trained classifiers are restricted to very small models, limited training set sizes, and low-resolution images. Such restrictions hinder their applicability to real-world scenarios. In this paper, we present a novel approach enabling data reconstruction in realistic settings for models trained on high-resolution images. Our method adapts the reconstruction scheme of arXiv:2206.07758 to real-world scenarios -- specifically, targeting models trained via transfer learning over image embeddings of large pre-trained models like DINO-ViT and CLIP. Our work employs data reconstruction in the embedding space rather than in the image space, showcasing its applicability beyond visual data. Moreover, we introduce a novel clustering-based method to identify good reconstructions from thousands of candidates. This significantly improves on previous works that relied on knowledge of the training set to identify good reconstructed images. Our findings shed light on a potential privacy risk for data leakage from models trained using transfer learning.
Authors: Ziliang Zhang
Abstract: Emerging research in edge devices and micro-controller units (MCU) enables on-device computation of Deep Learning Training and Inferencing tasks. More recently, contemporary trends focus on making the Deep Neural Net (DNN) Models runnable on battery-less intermittent devices. One of the approaches is to shrink the DNN models by enabling weight sharing, pruning, and conducted Neural Architecture Search (NAS) with optimized search space to target specific edge devices \cite{Cai2019OnceFA} \cite{Lin2020MCUNetTD} \cite{Lin2021MCUNetV2MP} \cite{Lin2022OnDeviceTU}. Another approach analyzes the intermittent execution and designs the corresponding system by performing NAS that is aware of intermittent execution cycles and resource constraints \cite{iNAS} \cite{HW-NAS} \cite{iLearn}. However, the optimized NAS was only considering consecutive execution with no power loss, and intermittent execution designs only focused on balancing data reuse and costs related to intermittent inference and often with low accuracy. We proposed Accelerated Intermittent Deep Inference to harness the power of optimized inferencing DNN models specifically targeting SRAM under 256KB and make it schedulable and runnable within intermittent power. Our main contribution is: (1) Schedule tasks performed by on-device inferencing into intermittent execution cycles and optimize for latency; (2) Develop a system that can satisfy the end-to-end latency while achieving a much higher accuracy compared to baseline \cite{iNAS} \cite{HW-NAS}
Authors: Michael Beukman, Branden Ingram, Geraud Nangue Tasse, Benjamin Rosman, Pravesh Ranchod
Abstract: Reinforcement learning (RL) has progressed substantially over the past decade, with much of this progress being driven by benchmarks. Many benchmarks are focused on video or board games, and a large number of robotics benchmarks lack diversity and real-world applicability. In this paper, we aim to simplify the process of applying reinforcement learning in the 3D simulation league of Robocup, a robotic football competition. To this end, we introduce a Robocup-based RL environment based on the open source rcssserver3d soccer server, simple pre-defined tasks, and integration with a popular RL library, Stable Baselines 3. Our environment enables the creation of high-dimensional continuous control tasks within a robotics football simulation. In each task, an RL agent controls a simulated Nao robot, and can interact with the ball or other agents. We open-source our environment and training code at https://github.com/Michael-Beukman/RobocupGym.
Authors: Ritabrata Roy Choudhury
Abstract: The proposed model aims to develop a speech recognition technology for hearing, speech, or cognitively disabled people. All the available technology in the field of speech recognition doesn't come with an interface for communication for people with hearing, speech, or cognitive disabilities. The proposed model proposes the speech from the user, is transmitted to the speech recognition layer where it is converted into text and then that text is then transmitted to the morse code conversion layer where the morse code of the corresponding speech is given as the output. The accuracy of the model is completely dependent on speech recognition, as the morse code conversion is a process. The model is tested with recorded audio files with different parameters. The proposed model's WER and accuracy are both determined to be 10.18% and 89.82%, respectively.
Authors: Yongqian Sun, Jiaju Wang, Zhengdan Li, Xiaohui Nie, Minghua Ma, Shenglin Zhang, Yuhe Ji, Lu Zhang, Wen Long, Hengmao Chen, Yongnan Luo, Dan Pei
Abstract: AIOps algorithms play a crucial role in the maintenance of microservice systems. Many previous benchmarks' performance leaderboard provides valuable guidance for selecting appropriate algorithms. However, existing AIOps benchmarks mainly utilize offline datasets to evaluate algorithms. They cannot consistently evaluate the performance of algorithms using real-time datasets, and the operation scenarios for evaluation are static, which is insufficient for effective algorithm selection. To address these issues, we propose an evaluation-consistent and scenario-oriented evaluation framework named MicroServo. The core idea is to build a live microservice benchmark to generate real-time datasets and consistently simulate the specific operation scenarios on it. MicroServo supports different leaderboards by selecting specific algorithms and datasets according to the operation scenarios. It also supports the deployment of various types of algorithms, enabling algorithms hot-plugging. At last, we test MicroServo with three typical microservice operation scenarios to demonstrate its efficiency and usability.
Authors: Feilong Wang, Xuegang Ban, Peng Chen, Chenxi Liu, Rong Zhao
Abstract: Big mobility datasets (BMD) have shown many advantages in studying human mobility and evaluating the performance of transportation systems. However, the quality of BMD remains poorly understood. This study evaluates biases in BMD and develops mitigation methods. Using Google and Apple mobility data as examples, this study compares them with benchmark data from governmental agencies. Spatio-temporal discrepancies between BMD and benchmark are observed and their impacts on transportation applications are investigated, emphasizing the urgent need to address these biases to prevent misguided policymaking. This study further proposes and tests a bias mitigation method. It is shown that the mitigated BMD could generate valuable insights into large-scale public transit systems across 100+ US counties, revealing regional disparities of the recovery of transit systems from the COVID-19. This study underscores the importance of caution when using BMD in transportation research and presents effective mitigation strategies that would benefit practitioners.
Authors: Micha{\l} Kozielski, Marek Sikora, {\L}ukasz Wawrowski
Abstract: Rule-based models offer a human-understandable representation, i.e. they are interpretable. For this reason, they are used to explain the decisions of non-interpretable complex models, referred to as black box models. The generation of such explanations involves the approximation of a black box model by a rule-based model. To date, however, it has not been investigated whether the rule-based model makes decisions in the same way as the black box model it approximates. Decision making in the same way is understood in this work as the consistency of decisions and the consistency of the most important attributes used for decision making. This study proposes a novel approach ensuring that the rule-based surrogate model mimics the performance of the black box model. The proposed solution performs an explanation fusion involving rule generation and taking into account the feature importance determined by the selected XAI methods for the black box model being explained. The result of the method can be both global and local rule-based explanations. The quality of the proposed solution was verified by extensive analysis on 30 tabular benchmark datasets representing classification problems. Evaluation included comparison with the reference method and an illustrative case study. In addition, the paper discusses the possible pathways for the application of the rule-based approach in XAI and how rule-based explanations, including the proposed method, meet the user perspective and requirements for both content and presentation. The software created and a detailed report containing the full experimental results are available on the GitHub repository (https://github.com/ruleminer/FI-rules4XAI ).
Authors: Koosha Shirouyeh, Andrea Schiffauerova, Ashkan Ebadi
Abstract: Star scientists are highly influential researchers who have made significant contributions to their field, gained widespread recognition, and often attracted substantial research funding. They are critical for the advancement of science and innovation, and they have a significant influence on the transfer of knowledge and technology to industry. Identifying potential star scientists before their performance becomes outstanding is important for recruitment, collaboration, networking, or research funding decisions. Using machine learning techniques, this study proposes a model to predict star scientists in the field of artificial intelligence while highlighting features related to their success. Our results confirm that rising stars follow different patterns compared to their non-rising stars counterparts in almost all the early-career features. We also found that certain features such as gender and ethnic diversity play important roles in scientific collaboration and that they can significantly impact an author's career development and success. The most important features in predicting star scientists in the field of artificial intelligence were the number of articles, group discipline diversity, and weighted degree centrality. The proposed approach offers valuable insights for researchers, practitioners, and funding agencies interested in identifying and supporting talented researchers.
Authors: Yi Sheng, Hanchen Wang, Yipei Liu, Junhuan Yang, Weiwen Jiang, Youzuo Lin, Lei Yang
Abstract: Ultrasound computed tomography (USCT) is a promising technique that achieves superior medical imaging reconstruction resolution by fully leveraging waveform information, outperforming conventional ultrasound methods. Despite its advantages, high-quality USCT reconstruction relies on extensive data acquisition by a large number of transducers, leading to increased costs, computational demands, extended patient scanning times, and manufacturing complexities. To mitigate these issues, we propose a new USCT method called APS-USCT, which facilitates imaging with sparse data, substantially reducing dependence on high-cost dense data acquisition. Our APS-USCT method consists of two primary components: APS-wave and APS-FWI. The APS-wave component, an encoder-decoder system, preprocesses the waveform data, converting sparse data into dense waveforms to augment sample density prior to reconstruction. The APS-FWI component, utilizing the InversionNet, directly reconstructs the speed of sound (SOS) from the ultrasound waveform data. We further improve the model's performance by incorporating Squeeze-and-Excitation (SE) Blocks and source encoding techniques. Testing our method on a breast cancer dataset yielded promising results. It demonstrated outstanding performance with an average Structural Similarity Index (SSIM) of 0.8431. Notably, over 82% of samples achieved an SSIM above 0.8, with nearly 61% exceeding 0.85, highlighting the significant potential of our approach in improving USCT image reconstruction by efficiently utilizing sparse data.
Authors: Feiyang Li, Zinan Cao, Qixuan Yu, Yulu Gong, Xirui Tang
Abstract: For this paper, a prediction study of cloud computing energy consumption was conducted by optimising the data regression algorithm based on the horned lizard optimisation algorithm for Convolutional Neural Networks-Bi-Directional Gated Recurrent Units. Firstly, through Spearman correlation analysis of CPU, usage, memory usage, network traffic, power consumption, number of instructions executed, execution time and energy efficiency, we found that power consumption has the highest degree of positive correlation with energy efficiency, while CPU usage has the highest degree of negative correlation with energy efficiency. In our experiments, we introduced a random forest model and an optimisation model based on the horned lizard optimisation algorithm for testing, and the results show that the optimisation algorithm has better prediction results compared to the random forest model. Specifically, the mean square error (MSE) of the optimisation algorithm is 0.01 smaller than that of the random forest model, and the mean absolute error (MAE) is 0.01 smaller than that of the random forest.3 The results of the combined metrics show that the optimisation algorithm performs more accurately and reliably in predicting energy efficiency. This research result provides new ideas and methods to improve the energy efficiency of cloud computing systems. This research not only expands the scope of application in the field of cloud computing, but also provides a strong support for improving the energy use efficiency of the system.
Authors: Yuze Ge, Rujun Jiang
Abstract: The spectral risk has wide applications in machine learning, especially in real-world decision-making, where people are not only concerned with models' average performance. By assigning different weights to the losses of different sample points, rather than the same weights as in the empirical risk, it allows the model's performance to lie between the average performance and the worst-case performance. In this paper, we propose SOREL, the first stochastic gradient-based algorithm with convergence guarantees for the spectral risk minimization. Previous algorithms often consider adding a strongly concave function to smooth the spectral risk, thus lacking convergence guarantees for the original spectral risk. We theoretically prove that our algorithm achieves a near-optimal rate of $\widetilde{O}(1/\sqrt{\epsilon})$ in terms of $\epsilon$. Experiments on real datasets show that our algorithm outperforms existing algorithms in most cases, both in terms of runtime and sample complexity.
Authors: Srivishnu Vusirikala, Suraj Rajendran
Abstract: We investigate the potential of self-supervision in improving the accuracy of deep learning models trained to classify melanoma patches. Various self-supervision techniques such as rotation prediction, missing patch prediction, and corruption removal were implemented and assessed for their impact on the convolutional neural network's performance. Preliminary results suggest a positive influence of self-supervision methods on the model's accuracy. The study notably demonstrates the efficacy of the corruption removal method in enhancing model performance. Despite observable improvements, we conclude that the self-supervised models have considerable potential for further enhancement, achievable through training over more epochs or expanding the dataset. We suggest exploring other self-supervision methods like Bootstrap Your Own Latent (BYOL) and contrastive learning in future research, emphasizing the cost-benefit trade-off due to their resource-intensive nature. The findings underline the promise of self-supervision in augmenting melanoma detection capabilities of deep learning models.
Authors: Joyjit Kundu, Wenzhe Guo, Ali BanaGozar, Udari De Alwis, Sourav Sengupta, Puneet Gupta, Arindam Mallik
Abstract: Aligning future system design with the ever-increasing compute needs of large language models (LLMs) is undoubtedly an important problem in today's world. Here, we propose a general performance modeling methodology and workload analysis of distributed LLM training and inference through an analytical framework that accurately considers compute, memory sub-system, network, and various parallelization strategies (model parallel, data parallel, pipeline parallel, and sequence parallel). We validate our performance predictions with published data from literature and relevant industry vendors (e.g., NVIDIA). For distributed training, we investigate the memory footprint of LLMs for different activation re-computation methods, dissect the key factors behind the massive performance gain from A100 to B200 ($\sim$ 35x speed-up closely following NVIDIA's scaling trend), and further run a design space exploration at different technology nodes (12 nm to 1 nm) to study the impact of logic, memory, and network scaling on the performance. For inference, we analyze the compute versus memory boundedness of different operations at a matrix-multiply level for different GPU systems and further explore the impact of DRAM memory technology scaling on inference latency. Utilizing our modeling framework, we reveal the evolution of performance bottlenecks for both LLM training and inference with technology scaling, thus, providing insights to design future systems for LLM training and inference.
Authors: Martin Wattenberg, Fernanda B. Vi\'egas
Abstract: Many neural nets appear to represent data as linear combinations of "feature vectors." Algorithms for discovering these vectors have seen impressive recent success. However, we argue that this success is incomplete without an understanding of relational composition: how (or whether) neural nets combine feature vectors to represent more complicated relationships. To facilitate research in this area, this paper offers a guided tour of various relational mechanisms that have been proposed, along with preliminary analysis of how such mechanisms might affect the search for interpretable features. We end with a series of promising areas for empirical research, which may help determine how neural networks represent structured data.
Authors: Yizi Zhang, Yanchen Wang, Donato Jimenez-Beneto, Zixuan Wang, Mehdi Azabou, Blake Richards, Olivier Winter, The International Brain Laboratory, Eva Dyer, Liam Paninski, Cole Hurwitz
Abstract: Neuroscience research has made immense progress over the last decade, but our understanding of the brain remains fragmented and piecemeal: the dream of probing an arbitrary brain region and automatically reading out the information encoded in its neural activity remains out of reach. In this work, we build towards a first foundation model for neural spiking data that can solve a diverse set of tasks across multiple brain areas. We introduce a novel self-supervised modeling approach for population activity in which the model alternates between masking out and reconstructing neural activity across different time steps, neurons, and brain regions. To evaluate our approach, we design unsupervised and supervised prediction tasks using the International Brain Laboratory repeated site dataset, which is comprised of Neuropixels recordings targeting the same brain locations across 48 animals and experimental sessions. The prediction tasks include single-neuron and region-level activity prediction, forward prediction, and behavior decoding. We demonstrate that our multi-task-masking (MtM) approach significantly improves the performance of current state-of-the-art population models and enables multi-task learning. We also show that by training on multiple animals, we can improve the generalization ability of the model to unseen animals, paving the way for a foundation model of the brain at single-cell, single-spike resolution.
Authors: Saurav Muralidharan, Sharath Turuvekere Sreenivas, Raviraj Joshi, Marcin Chochowski, Mostofa Patwary, Mohammad Shoeybi, Bryan Catanzaro, Jan Kautz, Pavlo Molchanov
Abstract: Large language models (LLMs) targeting different deployment scales and sizes are currently produced by training each variant from scratch; this is extremely compute-intensive. In this paper, we investigate if pruning an existing LLM and then re-training it with a fraction (<3%) of the original training data can be a suitable alternative to repeated, full retraining. To this end, we develop a set of practical and effective compression best practices for LLMs that combine depth, width, attention and MLP pruning with knowledge distillation-based retraining; we arrive at these best practices through a detailed empirical exploration of pruning strategies for each axis, methods to combine axes, distillation strategies, and search techniques for arriving at optimal compressed architectures. We use this guide to compress the Nemotron-4 family of LLMs by a factor of 2-4x, and compare their performance to similarly-sized models on a variety of language modeling tasks. Deriving 8B and 4B models from an already pretrained 15B model using our approach requires up to 40x fewer training tokens per model compared to training from scratch; this results in compute cost savings of 1.8x for training the full model family (15B, 8B, and 4B). Minitron models exhibit up to a 16% improvement in MMLU scores compared to training from scratch, perform comparably to other community models such as Mistral 7B, Gemma 7B and Llama-3 8B, and outperform state-of-the-art compression techniques from the literature. We have open-sourced Minitron model weights on Huggingface, with corresponding supplementary material including example code available on GitHub.
Authors: Martin E. Malandro
Abstract: We introduce Composer's Assistant 2, a system for interactive human-computer composition in the REAPER digital audio workstation. Our work upgrades the Composer's Assistant system (which performs multi-track infilling of symbolic music at the track-measure level) with a wide range of new controls to give users fine-grained control over the system's outputs. Controls introduced in this work include two types of rhythmic conditioning controls, horizontal and vertical note onset density controls, several types of pitch controls, and a rhythmic interest control. We train a T5-like transformer model to implement these controls and to serve as the backbone of our system. With these controls, we achieve a dramatic improvement in objective metrics over the original system. We also study how well our model understands the meaning of our controls, and we conduct a listening study that does not find a significant difference between real music and music composed in a co-creative fashion with our system. We release our complete system, consisting of source code, pretrained models, and REAPER scripts.
Authors: Manuel Eberhardinger, Florian Rupp, Johannes Maucher, Setareh Maghsudi
Abstract: Despite tremendous progress, machine learning and deep learning still suffer from incomprehensible predictions. Incomprehensibility, however, is not an option for the use of (deep) reinforcement learning in the real world, as unpredictable actions can seriously harm the involved individuals. In this work, we propose a genetic programming framework to generate explanations for the decision-making process of already trained agents by imitating them with programs. Programs are interpretable and can be executed to generate explanations of why the agent chooses a particular action. Furthermore, we conduct an ablation study that investigates how extending the domain-specific language by using library learning alters the performance of the method. We compare our results with the previous state of the art for this problem and show that we are comparable in performance but require much less hardware resources and computation time.
Authors: Cuong Pham, Hoang Anh Dung, Cuong C. Nguyen, Trung Le, Dinh Phung, Gustavo Carneiro, Thanh-Toan Do
Abstract: Post-Training Quantization (PTQ) has received significant attention because it requires only a small set of calibration data to quantize a full-precision model, which is more practical in real-world applications in which full access to a large training set is not available. However, it often leads to overfitting on the small calibration dataset. Several methods have been proposed to address this issue, yet they still rely on only the calibration set for the quantization and they do not validate the quantized model due to the lack of a validation set. In this work, we propose a novel meta-learning based approach to enhance the performance of post-training quantization. Specifically, to mitigate the overfitting problem, instead of only training the quantized model using the original calibration set without any validation during the learning process as in previous PTQ works, in our approach, we both train and validate the quantized model using two different sets of images. In particular, we propose a meta-learning based approach to jointly optimize a transformation network and a quantized model through bi-level optimization. The transformation network modifies the original calibration data and the modified data will be used as the training set to learn the quantized model with the objective that the quantized model achieves a good performance on the original calibration data. Extensive experiments on the widely used ImageNet dataset with different neural network architectures demonstrate that our approach outperforms the state-of-the-art PTQ methods.
Authors: Angelly Cabrera, Kleanthis Avramidis, Shrikanth Narayanan
Abstract: Coffee leaf rust, a foliar disease caused by the fungus Hemileia vastatrix, poses a major threat to coffee production, especially in Central America. Climate change further aggravates this issue, as it shortens the latency period between initial infection and the emergence of visible symptoms in diseases like leaf rust. Shortened latency periods can lead to more severe plant epidemics and faster spread of diseases. There is, hence, an urgent need for effective disease management strategies. To address these challenges, we explore the potential of deep learning models for enhancing early disease detection. However, deep learning models require extensive processing power and large amounts of data for model training, resources that are typically scarce. To overcome these barriers, we propose a preprocessing technique that involves convolving training images with a high-pass filter to enhance lesion-leaf contrast, significantly improving model efficacy in resource-limited environments. This method and our model demonstrated a strong performance, achieving over 90% across all evaluation metrics--including precision, recall, F1-score, and the Dice coefficient. Our experiments show that this approach outperforms other methods, including two different image preprocessing techniques and using unaltered, full-color images.
Authors: Yukai Shi, Zhipeng Weng, Yupei Lin, Cidan Shi, Xiaojun Yang, Liang Lin
Abstract: In recent years, as computer vision tasks have increasingly relied on high-quality image inputs, the task of image dehazing has received significant attention. Previously, many methods based on priors and deep learning have been proposed to address the task of image dehazing. Ignoring the domain gap between different data, former de-hazing methods usually adopt multiple datasets for explicit training, which often makes the methods themselves be violated. To address this problem, we propose a novel method of internal and external data augmentation to improve the existing dehazing methodology. By using cross-data external augmentor. The dataset inherits samples from different domains that are firmly aligned, making the model learn more robust and generalizable features. By using the internal data augmentation method, the model can fully exploit local information within the images, thereby obtaining more image details. To demonstrate the effectiveness of our proposed method, we conduct training on both the Natural Image Dataset (NID) and the Remote Sensing Image Dataset (RSID). Experimental results show that our method clearly resolves the domain gap in different dehazing datasets and presents a new pipeline for joint training in the dehazing task. Our approach significantly outperforms other advanced methods in dehazing and produces dehazed images that are closest to real haze-free images. The code will be available at: https://github.com/wengzp1/ScaleUpDehazing
Authors: Christos Kyrkou
Abstract: High-efficiency deep learning (DL) models are necessary not only to facilitate their use in devices with limited resources but also to improve resources required for training. Convolutional neural networks (ConvNets) typically exert severe demands on local device resources and this conventionally limits their adoption within mobile and embedded platforms. This brief presents work toward utilizing static convolutional filters generated from the space of local binary patterns (LBPs) and Haar features to design efficient ConvNet architectures. These are referred to as Structured Ternary Patterns (STePs) and can be generated during network initialization in a systematic way instead of having learnable weight parameters thus reducing the total weight updates. The ternary values require significantly less storage and with the appropriate low-level implementation, can also lead to inference improvements. The proposed approach is validated using four image classification datasets, demonstrating that common network backbones can be made more efficient and provide competitive results. It is also demonstrated that it is possible to generate completely custom STeP-based networks that provide good trade-offs for on-device applications such as unmanned aerial vehicle (UAV)-based aerial vehicle detection. The experimental results show that the proposed method maintains high detection accuracy while reducing the trainable parameters by 40-80%. This work motivates further research toward good priors for non-learnable weights that can make DL architectures more efficient without having to alter the network during or after training.
Authors: Alexandre Abraham, Andr\'es Hoyos Idrobo
Abstract: With the growing access to administrative health databases, retrospective studies have become crucial evidence for medical treatments. Yet, non-randomized studies frequently face selection biases, requiring mitigation strategies. Propensity score matching (PSM) addresses these biases by selecting comparable populations, allowing for analysis without further methodological constraints. However, PSM has several drawbacks. Different matching methods can produce significantly different Average Treatment Effects (ATE) for the same task, even when meeting all validation criteria. To prevent cherry-picking the best method, public authorities must involve field experts and engage in extensive discussions with researchers. To address this issue, we introduce a novel metric, A2A, to reduce the number of valid matches. A2A constructs artificial matching tasks that mirror the original ones but with known outcomes, assessing each matching method's performance comprehensively from propensity estimation to ATE estimation. When combined with Standardized Mean Difference, A2A enhances the precision of model selection, resulting in a reduction of up to 50% in ATE estimation errors across synthetic tasks and up to 90% in predicted ATE variability across both synthetic and real-world datasets. To our knowledge, A2A is the first metric capable of evaluating outcome correction accuracy using covariates not involved in selection. Computing A2A requires solving hundreds of PSMs, we therefore automate all manual steps of the PSM pipeline. We integrate PSM methods from Python and R, our automated pipeline, a new metric, and reproducible experiments into popmatch, our new Python package, to enhance reproducibility and accessibility to bias correction methods.
Authors: Francesco Di Luzio, Antonello Rosato, Massimo Panella
Abstract: In the context of artificial intelligence, the inherent human attribute of engaging in logical reasoning to facilitate decision-making is mirrored by the concept of explainability, which pertains to the ability of a model to provide a clear and interpretable account of how it arrived at a particular outcome. This study explores explainability techniques for binary deep neural architectures in the framework of emotion classification through video analysis. We investigate the optimization of input features to binary classifiers for emotion recognition, with face landmarks detection using an improved version of the Integrated Gradients explainability method. The main contribution of this paper consists in the employment of an innovative explainable artificial intelligence algorithm to understand the crucial facial landmarks movements during emotional feeling, using this information also for improving the performances of deep learning-based emotion classifiers. By means of explainability, we can optimize the number and the position of the facial landmarks used as input features for facial emotion recognition, lowering the impact of noisy landmarks and thus increasing the accuracy of the developed models. In order to test the effectiveness of the proposed approach, we considered a set of deep binary models for emotion classification trained initially with a complete set of facial landmarks, which are progressively reduced based on a suitable optimization procedure. The obtained results prove the robustness of the proposed explainable approach in terms of understanding the relevance of the different facial points for the different emotions, also improving the classification accuracy and diminishing the computational cost.
Authors: Silviu Pitis, Ziang Xiao, Nicolas Le Roux, Alessandro Sordoni
Abstract: While finetuning language models from pairwise preferences has proven remarkably effective, the underspecified nature of natural language presents critical challenges. Direct preference feedback is uninterpretable, difficult to provide where multidimensional criteria may apply, and often inconsistent, either because it is based on incomplete instructions or provided by diverse principals. To address these challenges, we consider the two-step preference modeling procedure that first resolves the under-specification by selecting a context, and then evaluates preference with respect to the chosen context. We decompose reward modeling error according to these two steps, which suggests that supervising context in addition to context-specific preference may be a viable approach to aligning models with diverse human preferences. For this to work, the ability of models to evaluate context-specific preference is critical. To this end, we contribute context-conditioned preference datasets and accompanying experiments that investigate the ability of language models to evaluate context-specific preference. We use our datasets to (1) show that existing preference models benefit from, but fail to fully consider, added context, (2) finetune a context-aware reward model with context-specific performance exceeding that of GPT-4 and Llama 3 70B on tested datasets, and (3) investigate the value of context-aware preference modeling.
Authors: Shayne Longpre, Robert Mahari, Ariel Lee, Campbell Lund, Hamidah Oderinwale, William Brannon, Nayan Saxena, Naana Obeng-Marnu, Tobin South, Cole Hunter, Kevin Klyman, Christopher Klamm, Hailey Schoelkopf, Nikhil Singh, Manuel Cherep, Ahmad Anis, An Dinh, Caroline Chitongo, Da Yin, Damien Sileo, Deividas Mataciunas, Diganta Misra, Emad Alghamdi, Enrico Shippole, Jianguo Zhang, Joanna Materzynska, Kun Qian, Kush Tiwary, Lester Miranda, Manan Dey, Minnie Liang, Mohammed Hamdy, Niklas Muennighoff, Seonghyeon Ye, Seungone Kim, Shrestha Mohanty, Vipul Gupta, Vivek Sharma, Vu Minh Chien, Xuhui Zhou, Yizhi Li, Caiming Xiong, Luis Villa, Stella Biderman, Hanlin Li, Daphne Ippolito, Sara Hooker, Jad Kabbara, Sandy Pentland
Abstract: General-purpose artificial intelligence (AI) systems are built on massive swathes of public web data, assembled into corpora such as C4, RefinedWeb, and Dolma. To our knowledge, we conduct the first, large-scale, longitudinal audit of the consent protocols for the web domains underlying AI training corpora. Our audit of 14,000 web domains provides an expansive view of crawlable web data and how consent preferences to use it are changing over time. We observe a proliferation of AI-specific clauses to limit use, acute differences in restrictions on AI developers, as well as general inconsistencies between websites' expressed intentions in their Terms of Service and their robots.txt. We diagnose these as symptoms of ineffective web protocols, not designed to cope with the widespread re-purposing of the internet for AI. Our longitudinal analyses show that in a single year (2023-2024) there has been a rapid crescendo of data restrictions from web sources, rendering ~5%+ of all tokens in C4, or 28%+ of the most actively maintained, critical sources in C4, fully restricted from use. For Terms of Service crawling restrictions, a full 45% of C4 is now restricted. If respected or enforced, these restrictions are rapidly biasing the diversity, freshness, and scaling laws for general-purpose AI systems. We hope to illustrate the emerging crisis in data consent, foreclosing much of the open web, not only for commercial AI, but non-commercial AI and academic purposes.
Authors: Dmitrii Galimzianov, Viacheslav Vyshegorodtsev
Abstract: Interruption in a dialogue occurs when the listener begins their speech before the current speaker finishes speaking. Interruptions can be broadly divided into two groups: cooperative (when the listener wants to support the speaker), and competitive (when the listener tries to take control of the conversation against the speaker's will). A system that automatically classifies interruptions can be used in call centers, specifically in the tasks of customer satisfaction monitoring and agent monitoring. In this study, we developed a text-based interruption classification model by preparing an in-house dataset consisting of ASR-transcribed customer support telephone dialogues in Russian. We fine-tuned Conversational RuBERT on our dataset and optimized hyperparameters, and the model performed well. With further improvements, the proposed model can be applied to automatic monitoring systems.
Authors: Desta Haileselassie Hagos, Rick Battle, Danda B. Rawat
Abstract: The emergence of Generative Artificial Intelligence (AI) and Large Language Models (LLMs) has marked a new era of Natural Language Processing (NLP), introducing unprecedented capabilities that are revolutionizing various domains. This paper explores the current state of these cutting-edge technologies, demonstrating their remarkable advancements and wide-ranging applications. Our paper contributes to providing a holistic perspective on the technical foundations, practical applications, and emerging challenges within the evolving landscape of Generative AI and LLMs. We believe that understanding the generative capabilities of AI systems and the specific context of LLMs is crucial for researchers, practitioners, and policymakers to collaboratively shape the responsible and ethical integration of these technologies into various domains. Furthermore, we identify and address main research gaps, providing valuable insights to guide future research endeavors within the AI research community.
Authors: Md Laraib Salam, Akash S Balsaraf, Gaurav Gupta
Abstract: The use of neural networks and deep learning techniques in image processing has significantly advanced the field, enabling highly accurate recognition results. However, achieving high recognition rates often necessitates complex network models, which can be challenging to train and require substantial computational resources. This research presents a simplified yet effective approach to predicting both the base and exponent from images of mathematical expressions using a multi-output Convolutional Neural Network (CNN). The model is trained on 10,900 synthetically generated images containing exponent expressions, incorporating random noise, font size variations, and blur intensity to simulate real-world conditions. The proposed CNN model demonstrates robust performance with efficient training time. The experimental results indicate that the model achieves high accuracy in predicting the base and exponent values, proving the efficacy of this approach in handling noisy and varied input images.
Authors: Md Zarif Hossain, Ahmed Imteaj
Abstract: Vision-language models (VLMs) have achieved significant strides in recent times specially in multimodal tasks, yet they remain susceptible to adversarial attacks on their vision components. To address this, we propose Sim-CLIP, an unsupervised adversarial fine-tuning method that enhances the robustness of the widely-used CLIP vision encoder against such attacks while maintaining semantic richness and specificity. By employing a Siamese architecture with cosine similarity loss, Sim-CLIP learns semantically meaningful and attack-resilient visual representations without requiring large batch sizes or momentum encoders. Our results demonstrate that VLMs enhanced with Sim-CLIP's fine-tuned CLIP encoder exhibit significantly enhanced robustness against adversarial attacks, while preserving semantic meaning of the perturbed images. Notably, Sim-CLIP does not require additional training or fine-tuning of the VLM itself; replacing the original vision encoder with our fine-tuned Sim-CLIP suffices to provide robustness. This work underscores the significance of reinforcing foundational models like CLIP to safeguard the reliability of downstream VLM applications, paving the way for more secure and effective multimodal systems.
Authors: Antonis Antoniades, Xinyi Wang, Yanai Elazar, Alfonso Amayuelas, Alon Albalak, Kexun Zhang, William Yang Wang
Abstract: Despite the proven utility of large language models (LLMs) in real-world applications, there remains a lack of understanding regarding how they leverage their large-scale pretraining text corpora to achieve such capabilities. In this work, we investigate the interplay between generalization and memorization in pretrained LLMs at scale, through a comprehensive $n$-gram analysis of their training data. Our experiments focus on three general task types: translation, question-answering, and multiple-choice reasoning. With various sizes of open-source LLMs and their pretraining corpora, we observe that as the model size increases, the task-relevant $n$-gram pair data becomes increasingly important, leading to improved task performance, decreased memorization, stronger generalization, and emergent abilities. Our results support the hypothesis that LLMs' capabilities emerge from a delicate balance of memorization and generalization with sufficient task-related pretraining data, and point the way to larger-scale analyses that could further improve our understanding of these models.
Authors: Alissa A. Valentine, Alexander W. Charney, Isotta Landi
Abstract: As interest in implementing artificial intelligence (AI) in medical systems grows, discussion continues on how to evaluate the fairness of these systems, or the disparities they may perpetuate. Socioeconomic status (SES) is commonly included in machine learning models to control for health inequities, with the underlying assumption that increased SES is associated with better health. In this work, we considered a large cohort of patients from the Mount Sinai Health System in New York City to investigate the effect of patient SES, race, and sex on schizophrenia (SCZ) diagnosis rates via a logistic regression model. Within an intersectional framework, patient SES, race, and sex were found to have significant interactions. Our findings showed that increased SES is associated with a higher probability of obtaining a SCZ diagnosis in Black Americans ($\beta=4.1\times10^{-8}$, $SE=4.5\times10^{-9}$, $p < 0.001$). Whereas high SES acts as a protective factor for SCZ diagnosis in White Americans ($\beta=-4.1\times10^{-8}$, $SE=6.7\times10^{-9}$, $p < 0.001$). Further investigation is needed to reliably explain and quantify health disparities. Nevertheless, we advocate that building fair AI tools for the health care space requires recognizing the intersectionality of sociodemographic factors.
Authors: Fadel M. Megahed, Ying-Ju Chen, Joshua A. Ferris, Cameron Resatar, Kaitlyn Ross, Younghwa Lee, L. Allison Jones-Farmer
Abstract: As generative AI continues to evolve, educators face the challenge of preparing students for a future where AI-assisted work is integral to professional success. This paper introduces ChatISA, an in-house, multi-model chatbot designed to support students in an Information Systems and Analytics department. ChatISA comprises four primary modules-Coding Companion, Project Coach, Exam Ally, and Interview Mentor-each tailored to enhance different aspects of the educational experience. Through iterative development, student feedback, and leveraging open-source frameworks, we created a robust tool that addresses coding inquiries, project management, exam preparation, and interview readiness. The implementation of ChatISA revealed significant insights and challenges, including the necessity of ethical guidelines and balancing AI usage with maintaining student agency. Our findings underscore the importance of adaptive pedagogy and proactive engagement with AI tools to maximize their educational benefits. To support broader adoption and innovation, all code for ChatISA is made publicly available on GitHub, enabling other institutions to customize and integrate similar AI-driven educational tools within their curricula.
Authors: Masoud Bashiri, Kamran Kowsari
Abstract: The advent of Large Language Models (LLMs) and Artificial Intelligence (AI) tools has revolutionized various facets of our lives, particularly in the realm of social media. For students, these advancements have unlocked unprecedented opportunities for learning, collaboration, and personal growth. AI-driven applications are transforming how students interact with social media, offering personalized content and recommendations, and enabling smarter, more efficient communication. Recent studies utilizing data from UniversityCube underscore the profound impact of AI tools on students' academic and social experiences. These studies reveal that students engaging with AI-enhanced social media platforms report higher academic performance, enhanced critical thinking skills, and increased engagement in collaborative projects. Moreover, AI tools assist in filtering out distracting content, allowing students to concentrate more on educational materials and pertinent discussions. The integration of LLMs in social media has further facilitated improved peer-to-peer communication and mentorship opportunities. AI algorithms effectively match students based on shared academic interests and career goals, fostering a supportive and intellectually stimulating online community, thereby contributing to increased student satisfaction and retention rates. In this article, we delve into the data provided by UniversityCube to explore how LLMs and AI tools are specifically transforming social media for students. Through case studies and statistical analyses, we offer a comprehensive understanding of the educational and social benefits these technologies offer. Our exploration highlights the potential of AI-driven tools to create a more enriched, efficient, and supportive educational environment for students in the digital age.
Authors: Mengru Wang, Yunzhi Yao, Ziwen Xu, Shuofei Qiao, Shumin Deng, Peng Wang, Xiang Chen, Jia-Chen Gu, Yong Jiang, Pengjun Xie, Fei Huang, Huajun Chen, Ningyu Zhang
Abstract: Understanding knowledge mechanisms in Large Language Models (LLMs) is crucial for advancing towards trustworthy AGI. This paper reviews knowledge mechanism analysis from a novel taxonomy including knowledge utilization and evolution. Knowledge utilization delves into the mechanism of memorization, comprehension and application, and creation. Knowledge evolution focuses on the dynamic progression of knowledge within individual and group LLMs. Moreover, we discuss what knowledge LLMs have learned, the reasons for the fragility of parametric knowledge, and the potential dark knowledge (hypothesis) that will be challenging to address. We hope this work can help understand knowledge in LLMs and provide insights for future research.
Authors: Meng Cao, Philip I. Pavlik Jr., Wei Chu, Liang Zhang
Abstract: In category learning, a growing body of literature has increasingly focused on exploring the impacts of interleaving in contrast to blocking. The sequential attention hypothesis posits that interleaving draws attention to the differences between categories while blocking directs attention toward similarities within categories. Although a recent study underscores the joint influence of memory and attentional factors on sequencing effects, there remains a scarcity of effective computational models integrating both attentional and memory considerations to comprehensively understand the effect of training sequences on students' performance. This study introduces a novel integration of attentional factors and spacing into the logistic knowledge tracing (LKT) models to monitor students' performance across different training sequences (interleaving and blocking). Attentional factors were incorporated by recording the counts of comparisons between adjacent trials, considering whether they belong to the same or different category. Several features were employed to account for temporal spacing. We used cross-validations to test the model fit and predictions on the learning session and posttest. Our findings reveal that incorporating both attentional factors and spacing features in the Additive Factors Model (AFM) significantly enhances its capacity to capture the effects of interleaving and blocking and demonstrates superior predictive accuracy for students' learning outcomes. By bridging the gap between attentional factors and memory processes, our computational approach offers a more comprehensive framework for understanding and predicting category learning outcomes in educational settings.
Authors: Tao Li, Zilin Bian, Haozhe Lei, Fan Zuo, Ya-Ting Yang, Quanyan Zhu, Zhenning Li, Zhibin Chen, Kaan Ozbay
Abstract: Traditional mobility management strategies emphasize macro-level mobility oversight from traffic-sensing infrastructures, often overlooking safety risks that directly affect road users. To address this, we propose a Digital Twin-based Driver Risk-Aware Intelligent Mobility Analytics (DT-DIMA) system. The DT-DIMA system integrates real-time traffic information from pan-tilt-cameras (PTCs), synchronizes this data into a digital twin to accurately replicate the physical world, and predicts network-wide mobility and safety risks in real time. The system's innovation lies in its integration of spatial-temporal modeling, simulation, and online control modules. Tested and evaluated under normal traffic conditions and incidental situations (e.g., unexpected accidents, pre-planned work zones) in a simulated testbed in Brooklyn, New York, DT-DIMA demonstrated mean absolute percentage errors (MAPEs) ranging from 8.40% to 15.11% in estimating network-level traffic volume and MAPEs from 0.85% to 12.97% in network-level safety risk prediction. In addition, the highly accurate safety risk prediction enables PTCs to preemptively monitor road segments with high driving risks before incidents take place. Such proactive PTC surveillance creates around a 5-minute lead time in capturing traffic incidents. The DT-DIMA system enables transportation managers to understand mobility not only in terms of traffic patterns but also driver-experienced safety risks, allowing for proactive resource allocation in response to various traffic situations. To the authors' best knowledge, DT-DIMA is the first urban mobility management system that considers both mobility and safety risks based on digital twin architecture.
Authors: Wangkun Xu, Qian Chen, Pudong Ge, Zhongda Chu, Fei Teng
Abstract: Encoding frequency stability constraints in the operation problem is challenging due to its complex dynamics. Recently, data-driven approaches have been proposed to learn the stability criteria offline with the trained model embedded as a constraint of online optimization. However, random sampling of stationary operation points is less efficient in generating balanced stable and unstable samples. Meanwhile, the performance of such a model is strongly dependent on the quality of the training dataset. Observing this research gap, we propose a gradient-based data generation method via forward-mode automatic differentiation. In this method, the original dynamic system is augmented with new states that represent the dynamic of sensitivities of the original states, which can be solved by invoking any ODE solver for a single time. To compensate for the contradiction between the gradient of various frequency stability criteria, gradient surgery is proposed by projecting the gradient on the normal plane of the other. In the end, we demonstrate the superior performance of the proposed sampling algorithm, compared with the unrolling differentiation and finite difference. All codes are available at https://github.com/xuwkk/frequency_sample_ad.
Authors: Pu Ren, Rie Nakata, Maxime Lacour, Ilan Naiman, Nori Nakata, Jialin Song, Zhengfa Bi, Osman Asif Malik, Dmitriy Morozov, Omri Azencot, N. Benjamin Erichson, Michael W. Mahoney
Abstract: Predicting high-fidelity ground motions for future earthquakes is crucial for seismic hazard assessment and infrastructure resilience. Conventional empirical simulations suffer from sparse sensor distribution and geographically localized earthquake locations, while physics-based methods are computationally intensive and require accurate representations of Earth structures and earthquake sources. We propose a novel artificial intelligence (AI) simulator, Conditional Generative Modeling for Ground Motion (CGM-GM), to synthesize high-frequency and spatially continuous earthquake ground motion waveforms. CGM-GM leverages earthquake magnitudes and geographic coordinates of earthquakes and sensors as inputs, learning complex wave physics and Earth heterogeneities, without explicit physics constraints. This is achieved through a probabilistic autoencoder that captures latent distributions in the time-frequency domain and variational sequential models for prior and posterior distributions. We evaluate the performance of CGM-GM using small-magnitude earthquake records from the San Francisco Bay Area, a region with high seismic risks. CGM-GM demonstrates a strong potential for outperforming a state-of-the-art non-ergodic empirical ground motion model and shows great promise in seismology and beyond.
Authors: Hao Li, Zheng Li, Siyuan Wu, Chengrui Hu, Yutong Ye, Min Zhang, Dengguo Feng, Yang Zhang
Abstract: Most existing membership inference attacks (MIAs) utilize metrics (e.g., loss) calculated on the model's final state, while recent advanced attacks leverage metrics computed at various stages, including both intermediate and final stages, throughout the model training. Nevertheless, these attacks often process multiple intermediate states of the metric independently, ignoring their time-dependent patterns. Consequently, they struggle to effectively distinguish between members and non-members who exhibit similar metric values, particularly resulting in a high false-positive rate. In this study, we delve deeper into the new membership signals in the black-box scenario. We identify a new, more integrated membership signal: the Pattern of Metric Sequence, derived from the various stages of model training. We contend that current signals provide only partial perspectives of this new signal: the new one encompasses both the model's multiple intermediate and final states, with a greater emphasis on temporal patterns among them. Building upon this signal, we introduce a novel attack method called Sequential-metric based Membership Inference Attack (SeqMIA). Specifically, we utilize knowledge distillation to obtain a set of distilled models representing various stages of the target model's training. We then assess multiple metrics on these distilled models in chronological order, creating distilled metric sequence. We finally integrate distilled multi-metric sequences as a sequential multiformat and employ an attention-based RNN attack model for inference. Empirical results show SeqMIA outperforms all baselines, especially can achieve an order of magnitude improvement in terms of TPR @ 0.1% FPR. Furthermore, we delve into the reasons why this signal contributes to SeqMIA's high attack performance, and assess various defense mechanisms against SeqMIA.
Authors: Zonghao Huang, Neil Zhenqiang Gong, Michael K. Reiter
Abstract: Auditing the use of data in training machine-learning (ML) models is an increasingly pressing challenge, as myriad ML practitioners routinely leverage the effort of content creators to train models without their permission. In this paper, we propose a general method to audit an ML model for the use of a data-owner's data in training, without prior knowledge of the ML task for which the data might be used. Our method leverages any existing black-box membership inference method, together with a sequential hypothesis test of our own design, to detect data use with a quantifiable, tunable false-detection rate. We show the effectiveness of our proposed framework by applying it to audit data use in two types of ML models, namely image classifiers and foundation models.
Authors: Aishwarya Jadhav, Ritam Dutt
Abstract: The task of searching through patent documents is crucial for chemical patent recommendation and retrieval. This can be enhanced by creating a patent knowledge base (ChemPatKB) to aid in prior art searches and to provide a platform for domain experts to explore new innovations in chemical compound synthesis and use-cases. An essential foundational component of this KB is the extraction of important reaction snippets from long patents documents which facilitates multiple downstream tasks such as reaction co-reference resolution and chemical entity role identification. In this work, we explore the problem of extracting reactions spans from chemical patents in order to create a reactions resource database. We formulate this task as a paragraph-level sequence tagging problem, where the system is required to return a sequence of paragraphs that contain a description of a reaction. We propose several approaches and modifications of the baseline models and study how different methods generalize across different domains of chemical patents.
Authors: Junyoung Park, Myeonggu Kang, Yunki Han, Yanggon Kim, Jaekang Shin, Lee-Sup Kim
Abstract: The attention mechanism in text generation is memory-bounded due to its sequential characteristics. Therefore, off-chip memory accesses should be minimized for faster execution. Although previous methods addressed this by pruning unimportant tokens, they fall short in selectively removing tokens with near-zero attention probabilities in each instance. Our method estimates the probability before the softmax function, effectively removing low probability tokens and achieving an 12.1x pruning ratio without fine-tuning. Additionally, we present a hardware design supporting seamless on-demand off-chip access. Our approach shows 2.6x reduced memory accesses, leading to an average 2.3x speedup and a 2.4x energy efficiency.
Authors: Ari Tchetchenian, Leo Zekelman, Yuqian Chen, Jarrett Rushmore, Fan Zhang, Edward H. Yeterian, Nikos Makris, Yogesh Rathi, Erik Meijering, Yang Song, Lauren J. O'Donnell
Abstract: Parcellation of human cerebellar pathways is essential for advancing our understanding of the human brain. Existing diffusion MRI tractography parcellation methods have been successful in defining major cerebellar fibre tracts, while relying solely on fibre tract structure. However, each fibre tract may relay information related to multiple cognitive and motor functions of the cerebellum. Hence, it may be beneficial for parcellation to consider the potential importance of the fibre tracts for individual motor and cognitive functional performance measures. In this work, we propose a multimodal data-driven method for cerebellar pathway parcellation, which incorporates both measures of microstructure and connectivity, and measures of individual functional performance. Our method involves first training a multitask deep network to predict various cognitive and motor measures from a set of fibre tract structural features. The importance of each structural feature for predicting each functional measure is then computed, resulting in a set of structure-function saliency values that are clustered to parcellate cerebellar pathways. We refer to our method as Deep Multimodal Saliency Parcellation (DeepMSP), as it computes the saliency of structural measures for predicting cognitive and motor functional performance, with these saliencies being applied to the task of parcellation. Applying DeepMSP we found that it was feasible to identify multiple cerebellar pathway parcels with unique structure-function saliency patterns that were stable across training folds.
Authors: Yu Zhang, Ruijie Yu, Kaipeng Zeng, Ding Li, Feng Zhu, Xiaokang Yang, Yaohui Jin, Yanyan Xu
Abstract: High-throughput reaction condition (RC) screening is fundamental to chemical synthesis. However, current RC screening suffers from laborious and costly trial-and-error workflows. Traditional computer-aided synthesis planning (CASP) tools fail to find suitable RCs due to data sparsity and inadequate reaction representations. Nowadays, large language models (LLMs) are capable of tackling chemistry-related problems, such as molecule design, and chemical logic Q\&A tasks. However, LLMs have not yet achieved accurate predictions of chemical reaction conditions. Here, we present MM-RCR, a text-augmented multimodal LLM that learns a unified reaction representation from SMILES, reaction graphs, and textual corpus for chemical reaction recommendation (RCR). To train MM-RCR, we construct 1.2 million pair-wised Q\&A instruction datasets. Our experimental results demonstrate that MM-RCR achieves state-of-the-art performance on two open benchmark datasets and exhibits strong generalization capabilities on out-of-domain (OOD) and High-Throughput Experimentation (HTE) datasets. MM-RCR has the potential to accelerate high-throughput condition screening in chemical synthesis.
Authors: Gilad Yehudai, Haim Kaplan, Asma Ghandeharioun, Mor Geva, Amir Globerson
Abstract: Large language models based on the transformer architectures can solve highly complex tasks. But are there simple tasks that such models cannot solve? Here we focus on very simple counting tasks, that involve counting how many times a token in the vocabulary have appeared in a string. We show that if the dimension of the transformer state is linear in the context length, this task can be solved. However, the solution we propose does not scale beyond this limit, and we provide theoretical arguments for why it is likely impossible for a size limited transformer to implement this task. Our empirical results demonstrate the same phase-transition in performance, as anticipated by the theoretical argument. Our results demonstrate the importance of understanding how transformers can solve simple tasks.
Authors: Qian Feng, Jianxiang Feng, Zhaopeng Chen, Rudolph Triebel, Alois Knoll
Abstract: Synthesizing diverse and accurate grasps with multi-fingered hands is an important yet challenging task in robotics. Previous efforts focusing on generative modeling have fallen short of precisely capturing the multi-modal, high-dimensional grasp distribution. To address this, we propose exploiting a special kind of Deep Generative Model (DGM) based on Normalizing Flows (NFs), an expressive model for learning complex probability distributions. Specifically, we first observed an encouraging improvement in diversity by directly applying a single conditional NFs (cNFs), dubbed FFHFlow-cnf, to learn a grasp distribution conditioned on the incomplete point cloud. However, we also recognized limited performance gains due to restricted expressivity in the latent space. This motivated us to develop a novel flow-based d Deep Latent Variable Model (DLVM), namely FFHFlow-lvm, which facilitates more reasonable latent features, leading to both diverse and accurate grasp synthesis for unseen objects. Unlike Variational Autoencoders (VAEs), the proposed DLVM counteracts typical pitfalls such as mode collapse and mis-specified priors by leveraging two cNFs for the prior and likelihood distributions, which are usually restricted to being isotropic Gaussian. Comprehensive experiments in simulation and real-robot scenarios demonstrate that our method generates more accurate and diverse grasps than the VAE baselines. Additionally, a run-time comparison is conducted to reveal its high potential for real-time applications.
Authors: Kai T\"opfer, Luis Itza Vazquez-Salazar, Markus Meuwly
Abstract: With the establishment of machine learning (ML) techniques in the scientific community, the construction of ML potential energy surfaces (ML-PES) has become a standard process in physics and chemistry. So far, improvements in the construction of ML-PES models have been conducted independently, creating an initial hurdle for new users to overcome and complicating the reproducibility of results. Aiming to reduce the bar for the extensive use of ML-PES, we introduce ${\it Asparagus}$, a software package encompassing the different parts into one coherent implementation that allows an autonomous, user-guided construction of ML-PES models. ${\it Asparagus}$ combines capabilities of initial data sampling with interfaces to ${\it ab initio}$ calculation programs, ML model training, as well as model evaluation and its application within other codes such as ASE or CHARMM. The functionalities of the code are illustrated in different examples, including the dynamics of small molecules, the representation of reactive potentials in organometallic compounds, and atom diffusion on periodic surface structures. The modular framework of ${\it Asparagus}$ is designed to allow simple implementations of further ML-related methods and models to provide constant user-friendly access to state-of-the-art ML techniques.
Authors: Ricardo Ferreira, Filipa Valdeira, Marta Guimar\~aes, Cl\'audia Soares
Abstract: With the increase in the number of active satellites and space debris in orbit, the problem of initial orbit determination (IOD) becomes increasingly important, demanding a high accuracy. Over the years, different approaches have been presented such as filtering methods (for example, Extended Kalman Filter), differential algebra or solving Lambert's problem. In this work, we consider a setting of three monostatic radars, where all available measurements are taken approximately at the same instant. This follows a similar setting as trilateration, a state-of-the-art approach, where each radar is able to obtain a single measurement of range and range-rate. Differently, and due to advances in Multiple-Input Multiple-Output (MIMO) radars, we assume that each location is able to obtain a larger set of range, angle and Doppler shift measurements. Thus, our method can be understood as an extension of trilateration leveraging more recent technology and incorporating additional data. We formulate the problem as a Maximum Likelihood Estimator (MLE), which for some number of observations is asymptotically unbiased and asymptotically efficient. Through numerical experiments, we demonstrate that our method attains the same accuracy as the trilateration method for the same number of measurements and offers an alternative and generalization, returning a more accurate estimation of the satellite's state vector, as the number of available measurements increases.
Authors: Qizhi Pei, Lijun Wu, Zhenyu He, Jinhua Zhu, Yingce Xia, Shufang Xie, Rui Yan
Abstract: Drug-Target binding Affinity (DTA) prediction is essential for drug discovery. Despite the application of deep learning methods to DTA prediction, the achieved accuracy remain suboptimal. In this work, inspired by the recent success of retrieval methods, we propose $k$NN-DTA, a non-parametric embedding-based retrieval method adopted on a pre-trained DTA prediction model, which can extend the power of the DTA model with no or negligible cost. Different from existing methods, we introduce two neighbor aggregation ways from both embedding space and label space that are integrated into a unified framework. Specifically, we propose a \emph{label aggregation} with \emph{pair-wise retrieval} and a \emph{representation aggregation} with \emph{point-wise retrieval} of the nearest neighbors. This method executes in the inference phase and can efficiently boost the DTA prediction performance with no training cost. In addition, we propose an extension, Ada-$k$NN-DTA, an instance-wise and adaptive aggregation with lightweight learning. Results on four benchmark datasets show that $k$NN-DTA brings significant improvements, outperforming previous state-of-the-art (SOTA) results, e.g, on BindingDB IC$_{50}$ and $K_i$ testbeds, $k$NN-DTA obtains new records of RMSE $\bf{0.684}$ and $\bf{0.750}$. The extended Ada-$k$NN-DTA further improves the performance to be $\bf{0.675}$ and $\bf{0.735}$ RMSE. These results strongly prove the effectiveness of our method. Results in other settings and comprehensive studies/analyses also show the great potential of our $k$NN-DTA approach.
Authors: Rylan Schaeffer, Dan Valentine, Luke Bailey, James Chua, Crist\'obal Eyzaguirre, Zane Durante, Joe Benton, Brando Miranda, Henry Sleight, John Hughes, Rajashree Agrawal, Mrinank Sharma, Scott Emmons, Sanmi Koyejo, Ethan Perez
Abstract: The integration of new modalities into frontier AI systems offers exciting capabilities, but also increases the possibility such systems can be adversarially manipulated in undesirable ways. In this work, we focus on a popular class of vision-language models (VLMs) that generate text outputs conditioned on visual and textual inputs. We conducted a large-scale empirical study to assess the transferability of gradient-based universal image "jailbreaks" using a diverse set of over 40 open-parameter VLMs, including 18 new VLMs that we publicly release. Overall, we find that transferable gradient-based image jailbreaks are extremely difficult to obtain. When an image jailbreak is optimized against a single VLM or against an ensemble of VLMs, the jailbreak successfully jailbreaks the attacked VLM(s), but exhibits little-to-no transfer to any other VLMs; transfer is not affected by whether the attacked and target VLMs possess matching vision backbones or language models, whether the language model underwent instruction-following and/or safety-alignment training, or many other factors. Only two settings display partially successful transfer: between identically-pretrained and identically-initialized VLMs with slightly different VLM training data, and between different training checkpoints of a single VLM. Leveraging these results, we then demonstrate that transfer can be significantly improved against a specific target VLM by attacking larger ensembles of "highly-similar" VLMs. These results stand in stark contrast to existing evidence of universal and transferable text jailbreaks against language models and transferable adversarial attacks against image classifiers, suggesting that VLMs may be more robust to gradient-based transfer attacks.
Authors: Yancheng Wang, Yingzhen Yang
Abstract: Self-attention and transformers have been widely used in deep learning. Recent efforts have been devoted to incorporating transformer blocks into different neural architectures, including those with convolutions, leading to various visual transformers for computer vision tasks. In this paper, we propose a novel and compact transformer block, Transformer with Learnable Token Merging (LTM), or LTM-Transformer. LTM-Transformer performs token merging in a learnable scheme. LTM-Transformer is compatible with many popular and compact transformer networks, and it reduces the FLOPs and the inference time of the visual transformers while maintaining or even improving the prediction accuracy. In the experiments, we replace all the transformer blocks in popular visual transformers, including MobileViT, EfficientViT, ViT-S/16, and Swin-T, with LTM-Transformer blocks, leading to LTM-Transformer networks with different backbones. The LTM-Transformer is motivated by reduction of Information Bottleneck, and a novel and separable variational upper bound for the IB loss is derived. The architecture of mask module in our LTM blocks which generate the token merging mask is designed to reduce the derived upper bound for the IB loss. Extensive results on computer vision tasks evidence that LTM-Transformer renders compact and efficient visual transformers with comparable or much better prediction accuracy than the original visual transformers. The code of the LTM-Transformer is available at \url{https://github.com/Statistical-Deep-Learning/LTM}.
Authors: Yuliya Burankova, Miriam Abele, Mohammad Bakhtiari, Christine von T\"orne, Teresa Barth, Lisa Schweizer, Pieter Giesbertz, Johannes R. Schmidt, Stefan Kalkhof, Janina M\"uller-Deile, Peter A van Veelen, Yassene Mohammed, Elke Hammer, Lis Arend, Klaudia Adamowicz, Tanja Laske, Anne Hartebrodt, Tobias Frisch, Chen Meng, Julian Matschinske, Julian Sp\"ath, Richard R\"ottger, Veit Schw\"ammle, Stefanie M. Hauck, Stefan Lichtenthaler, Axel Imhof, Matthias Mann, Christina Ludwig, Bernhard Kuster, Jan Baumbach, Olga Zolotareva
Abstract: Quantitative mass spectrometry has revolutionized proteomics by enabling simultaneous quantification of thousands of proteins. Pooling patient-derived data from multiple institutions enhances statistical power but raises significant privacy concerns. Here we introduce FedProt, the first privacy-preserving tool for collaborative differential protein abundance analysis of distributed data, which utilizes federated learning and additive secret sharing. In the absence of a multicenter patient-derived dataset for evaluation, we created two, one at five centers from LFQ E.coli experiments and one at three centers from TMT human serum. Evaluations using these datasets confirm that FedProt achieves accuracy equivalent to DEqMS applied to pooled data, with completely negligible absolute differences no greater than $\text{$4 \times 10^{-12}$}$. In contrast, -log10(p-values) computed by the most accurate meta-analysis methods diverged from the centralized analysis results by up to 25-27. FedProt is available as a web tool with detailed documentation as a FeatureCloud App.
Authors: Jipeng Zhang, Yaxuan Qin, Renjie Pi, Weizhong Zhang, Rui Pan, Tong Zhang
Abstract: Instruction tuning has achieved unprecedented success in NLP, turning large language models into versatile chatbots. However, the increasing variety and volume of instruction datasets demand significant computational resources. To address this, it is essential to extract a small and highly informative subset (i.e., Coreset) that achieves comparable performance to the full dataset. Achieving this goal poses non-trivial challenges: 1) data selection requires accurate data representations that reflect the training samples' quality, 2) considering the diverse nature of instruction datasets, and 3) ensuring the efficiency of the coreset selection algorithm for large models. To address these challenges, we propose Task-Agnostic Gradient Clustered COreset Selection (TAGCOS). Specifically, we leverage sample gradients as the data representations, perform clustering to group similar data, and apply an efficient greedy algorithm for coreset selection. Experimental results show that our algorithm, selecting only 5% of the data, surpasses other unsupervised methods and achieves performance close to that of the full dataset.
Authors: Alexis M. H. Teter, Wenqing Wang, Abhishek Halder
Abstract: Schr\"{o}dinger bridge--a stochastic dynamical generalization of optimal mass transport--exhibits a learning-control duality. Viewed as a stochastic control problem, the Schr\"{o}dinger bridge finds an optimal control policy that steers a given joint state statistics to another while minimizing the total control effort subject to controlled diffusion and deadline constraints. Viewed as a stochastic learning problem, the Schr\"{o}dinger bridge finds the most-likely distribution-valued trajectory connecting endpoint distributional observations, i.e., solves the two point boundary-constrained maximum likelihood problem over the manifold of probability distributions. Recent works have shown that solving the Schr\"{o}dinger bridge problem with state cost requires finding the Markov kernel associated with a reaction-diffusion PDE where the state cost appears as a state-dependent reaction rate. We explain how ideas from Weyl calculus in quantum mechanics, specifically the Weyl operator and the Weyl symbol, can help determine such Markov kernels. We illustrate these ideas by explicitly finding the Markov kernel for the case of quadratic state cost via Weyl calculus, recovering our earlier results but avoiding tedious computation with Hermite polynomials.
Authors: Jeongmin Brian Park, Kun Wu, Vikram Sharma Mailthody, Zaid Quresh, Scott Mahlke, Wen-mei Hwu
Abstract: Graph Neural Networks (GNNs) are widely used today in recommendation systems, fraud detection, and node/link classification tasks. Real world GNNs continue to scale in size and require a large memory footprint for storing graphs and embeddings that often exceed the memory capacities of the target GPUs used for training. To address limited memory capacities, traditional GNN training approaches use graph partitioning and sharding techniques to scale up across multiple GPUs within a node and/or scale out across multiple nodes. However, this approach suffers from the high computational costs of graph partitioning algorithms and inefficient communication across GPUs. To address these overheads, we propose Large-scale Storage-based Multi-GPU GNN framework (LSM-GNN), a storagebased approach to train GNN models that utilizes a novel communication layer enabling GPU software caches to function as a system-wide shared cache with low overheads.LSM-GNN incorporates a hybrid eviction policy that intelligently manages cache space by using both static and dynamic node information to significantly enhance cache performance. Furthermore, we introduce the Preemptive Victim-buffer Prefetcher (PVP), a mechanism for prefetching node feature data from a Victim Buffer located in CPU pinned-memory to further reduce the pressure on the storage devices. Experimental results show that despite the lower compute capabilities and memory capacities, LSM-GNN in a single node with two GPUs offers superior performance over two-node-four-GPU Dist-DGL baseline and provides up to 3.75x speed up on end-to-end epoch time while running large-scale GNN training
Authors: Kwanyong Park, Kuniaki Saito, Donghyun Kim
Abstract: Vision-language (VL) models often exhibit a limited understanding of complex expressions of visual objects (e.g., attributes, shapes, and their relations), given complex and diverse language queries. Traditional approaches attempt to improve VL models using hard negative synthetic text, but their effectiveness is limited. In this paper, we harness the exceptional compositional understanding capabilities of generative foundational models. We introduce a novel method for structured synthetic data generation aimed at enhancing the compositional understanding of VL models in language-based object detection. Our framework generates densely paired positive and negative triplets (image, text descriptions, and bounding boxes) in both image and text domains. By leveraging these synthetic triplets, we transform 'weaker' VL models into 'stronger' models in terms of compositional understanding, a process we call "Weak-to-Strong Compositional Learning" (WSCL). To achieve this, we propose a new compositional contrastive learning formulation that discovers semantics and structures in complex descriptions from synthetic triplets. As a result, VL models trained with our synthetic data generation exhibit a significant performance boost in the Omnilabel benchmark by up to +5AP and the D3 benchmark by +6.9AP upon existing baselines.
Authors: Zhe Fei, Yi Li
Abstract: Epigenetic aging clocks play a pivotal role in estimating an individual's biological age through the examination of DNA methylation patterns at numerous CpG (Cytosine-phosphate-Guanine) sites within their genome. However, making valid inferences on predicted epigenetic ages, or more broadly, on predictions derived from high-dimensional inputs, presents challenges. We introduce a novel U-learning approach via combinatory multi-subsampling for making ensemble predictions and constructing confidence intervals for predictions of continuous outcomes when traditional asymptotic methods are not applicable. More specifically, our approach conceptualizes the ensemble estimators within the framework of generalized U-statistics and invokes the H\'ajek projection for deriving the variances of predictions and constructing confidence intervals with valid conditional coverage probabilities. We apply our approach to two commonly used predictive algorithms, Lasso and deep neural networks (DNNs), and illustrate the validity of inferences with extensive numerical studies. We have applied these methods to predict the DNA methylation age (DNAmAge) of patients with various health conditions, aiming to accurately characterize the aging process and potentially guide anti-aging interventions.
Authors: Jiale Xu, Rui Zhang, Cong Guo, Weiming Hu, Zihan Liu, Feiyang Wu, Yu Feng, Shixuan Sun, Changxu Shao, Yuhong Guo, Junping Zhao, Ke Zhang, Minyi Guo, Jingwen Leng
Abstract: Large Language Models (LLMs) are widely used across various domains, processing millions of daily requests. This surge in demand poses significant challenges in optimizing throughput and latency while keeping costs manageable. The Key-Value (KV) cache, a standard method for retaining previous computations, makes LLM inference highly bounded by memory. While batching strategies can enhance performance, they frequently lead to significant memory fragmentation. Even though cutting-edge systems like vLLM mitigate KV cache fragmentation using paged Attention mechanisms, they still suffer from inefficient memory and computational operations due to the tightly coupled page management and computation kernels. This study introduces the vTensor, an innovative tensor structure for LLM inference based on GPU virtual memory management (VMM). vTensor addresses existing limitations by decoupling computation from memory defragmentation and offering dynamic extensibility. Our framework employs a CPU-GPU heterogeneous approach, ensuring efficient, fragmentation-free memory management while accommodating various computation kernels across different LLM architectures. Experimental results indicate that vTensor achieves an average speedup of 1.86x across different models, with up to 2.42x in multi-turn chat scenarios. Additionally, vTensor provides average speedups of 2.12x and 3.15x in kernel evaluation, reaching up to 3.92x and 3.27x compared to SGLang Triton prefix-prefilling kernels and vLLM paged Attention kernel, respectively. Furthermore, it frees approximately 71.25% (57GB) of memory on the NVIDIA A100 GPU compared to vLLM, enabling more memory-intensive workloads.
Authors: Liekang Zeng (Sherman), Shengyuan Ye (Sherman), Xu Chen (Sherman), Xiaoxi Zhang (Sherman), Ju Ren (Sherman), Jian Tang (Sherman), Yang Yang (Sherman), Xuemin (Sherman), Shen
Abstract: Recent years have witnessed a thriving growth of computing facilities connected at the network edge, cultivating edge computing networks as a fundamental infrastructure for supporting miscellaneous intelligent services. Meanwhile, Artificial Intelligence frontiers have extrapolated Machine Learning to the graph domain and promoted Graph Intelligence (GI), which unlocks unprecedented ability in learning from massive data in graph structures. Given the inherent relation between graphs and networks, the interdiscipline of graph representation learning and edge networks, i.e., Edge GI or EGI, has revealed a novel interplay between them -- GI models principally open a new door for modeling, understanding, and optimizing edge networks, and conversely, edge networks serve as physical support for training, deploying, and accelerating GI models. Driven by this delicate closed-loop, EGI can be widely recognized as a promising solution to fully unleash the potential of edge computing power and is garnering significant attention. Nevertheless, research on EGI yet remains nascent, and there is a soaring demand within both the communications and AI communities for a dedicated venue to share recent advancements. To this end, this paper promotes the concept of EGI, explores its scope and core principles, and conducts a comprehensive survey concerning recent research efforts on this emerging field and specifically, introduces and discusses: 1) fundamentals of edge computing and graph representation learning, 2) emerging techniques centering on the closed loop between graph intelligence and edge networks, and 3) open challenges and research opportunities of future EGI. By bridging the gap across communication, networking, and graph learning areas, we believe that this survey can garner increased attention, foster meaningful discussions, and inspire further research ideas in EGI.
Authors: Zhao-Heng Yin, Pieter Abbeel
Abstract: Imitation learning is a powerful machine learning algorithm for a robot to acquire manipulation skills. Nevertheless, many real-world manipulation tasks involve precise and dexterous robot-object interactions, which make it difficult for humans to collect high-quality expert demonstrations. As a result, a robot has to learn skills from suboptimal demonstrations and unstructured interactions, which remains a key challenge. Existing works typically use offline deep reinforcement learning (RL) to solve this challenge, but in practice these algorithms are unstable and fragile due to the deadly triad issue. To overcome this problem, we propose GSR, a simple yet effective algorithm that learns from suboptimal demonstrations through Graph Search and Retrieval. We first use pretrained representation to organize the interaction experience into a graph and perform a graph search to calculate the values of different behaviors. Then, we apply a retrieval-based procedure to identify the best behavior (actions) on each state and use behavior cloning to learn that behavior. We evaluate our method in both simulation and real-world robotic manipulation tasks with complex visual inputs, covering various precise and dexterous manipulation skills with objects of different physical properties. GSR can achieve a 10% to 30% higher success rate and over 30% higher proficiency compared to baselines. Our project page is at https://zhaohengyin.github.io/gsr.
Authors: Huanjing Zhao, Beining Yang, Yukuo Cen, Junyu Ren, Chenhui Zhang, Yuxiao Dong, Evgeny Kharlamov, Shu Zhao, Jie Tang
Abstract: The text-attributed graph (TAG) is one kind of important real-world graph-structured data with each node associated with raw texts. For TAGs, traditional few-shot node classification methods directly conduct training on the pre-processed node features and do not consider the raw texts. The performance is highly dependent on the choice of the feature pre-processing method. In this paper, we propose P2TAG, a framework designed for few-shot node classification on TAGs with graph pre-training and prompting. P2TAG first pre-trains the language model (LM) and graph neural network (GNN) on TAGs with self-supervised loss. To fully utilize the ability of language models, we adapt the masked language modeling objective for our framework. The pre-trained model is then used for the few-shot node classification with a mixed prompt method, which simultaneously considers both text and graph information. We conduct experiments on six real-world TAGs, including paper citation networks and product co-purchasing networks. Experimental results demonstrate that our proposed framework outperforms existing graph few-shot learning methods on these datasets with +18.98% ~ +35.98% improvements.
Authors: Evgenii Chzhen (LMO, CELESTE), Mohamed Hebiri (LAMA), Gayane Taturyan (LAMA, IMT)
Abstract: We address the problem of performing regression while ensuring demographic parity, even without access to sensitive attributes during inference. We present a general-purpose post-processing algorithm that, using accurate estimates of the regression function and a sensitive attribute predictor, generates predictions that meet the demographic parity constraint. Our method involves discretization and stochastic minimization of a smooth convex function. It is suitable for online post-processing and multi-class classification tasks only involving unlabeled data for the post-processing. Unlike prior methods, our approach is fully theory-driven. We require precise control over the gradient norm of the convex function, and thus, we rely on more advanced techniques than standard stochastic gradient descent. Our algorithm is backed by finite-sample analysis and post-processing bounds, with experimental results validating our theoretical findings.
Authors: Elizabeth L. Baker, Moritz Schauer, Stefan Sommer
Abstract: We propose a new algorithm for learning a bridged diffusion process using score-matching methods. Our method relies on reversing the dynamics of the forward process and using this to learn a score function, which, via Doob's $h$-transform, gives us a bridged diffusion process; that is, a process conditioned on an endpoint. In contrast to prior methods, ours learns the score term $\nabla_x \log p(t, x; T, y)$, for given $t, Y$ directly, completely avoiding the need for first learning a time reversal. We compare the performance of our algorithm with existing methods and see that it outperforms using the (learned) time-reversals to learn the score term. The code can be found at https://github.com/libbylbaker/forward_bridge.
Authors: Bailu Ding, Jiaqi Zhai
Abstract: Retrieval plays a fundamental role in recommendation systems, search, and natural language processing by efficiently finding relevant items from a large corpus given a query. Dot products have been widely used as the similarity function in such retrieval tasks, thanks to Maximum Inner Product Search (MIPS) that enabled efficient retrieval based on dot products. However, state-of-the-art retrieval algorithms have migrated to learned similarities. Such algorithms vary in form; the queries can be represented with multiple embeddings, complex neural networks can be deployed, the item ids can be decoded directly from queries using beam search, and multiple approaches can be combined in hybrid solutions. Unfortunately, we lack efficient solutions for retrieval in these state-of-the-art setups. Our work investigates techniques for approximate nearest neighbor search with learned similarity functions. We first prove that Mixture-of-Logits (MoL) is a universal approximator, and can express all learned similarity functions. We next propose techniques to retrieve the approximate top K results using MoL with a tight bound. We finally compare our techniques with existing approaches, showing that MoL sets new state-of-the-art results on recommendation retrieval tasks, and our approximate top-k retrieval with learned similarities outperforms baselines by up to two orders of magnitude in latency, while achieving > .99 recall rate of exact algorithms.
Authors: \.Ismail \"Oz\c{c}\.il, A. Bu\u{g}ra Koku
Abstract: The advancement in computing power has significantly reduced the training times for deep learning, fostering the rapid development of networks designed for object recognition. However, the exploration of object utility, which is the affordance of the object, as opposed to object recognition, has received comparatively less attention. This work focuses on the problem of exploration of object affordances using existing networks trained on the object classification dataset. While pre-trained networks have proven to be instrumental in transfer learning for classification tasks, this work diverges from conventional object classification methods. Instead, it employs pre-trained networks to discern affordance labels without the need for specialized layers, abstaining from modifying the final layers through the addition of classification layers. To facilitate the determination of affordance labels without such modifications, two approaches, i.e. subspace clustering and manifold curvature methods are tested. These methods offer a distinct perspective on affordance label recognition. Especially, manifold curvature method has been successfully tested with nine distinct pre-trained networks, each achieving an accuracy exceeding 95%. Moreover, it is observed that manifold curvature and subspace clustering methods explore affordance labels that are not marked in the ground truth, but object affords in various cases.
Authors: Eugenio Lomurno, Matteo Matteucci
Abstract: Generative artificial intelligence has transformed the generation of synthetic data, providing innovative solutions to challenges like data scarcity and privacy, which are particularly critical in fields such as medicine. However, the effective use of this synthetic data to train high-performance models remains a significant challenge. This paper addresses this issue by introducing Knowledge Recycling (KR), a pipeline designed to optimise the generation and use of synthetic data for training downstream classifiers. At the heart of this pipeline is Generative Knowledge Distillation (GKD), the proposed technique that significantly improves the quality and usefulness of the information provided to classifiers through a synthetic dataset regeneration and soft labelling mechanism. The KR pipeline has been tested on a variety of datasets, with a focus on six highly heterogeneous medical image datasets, ranging from retinal images to organ scans. The results show a significant reduction in the performance gap between models trained on real and synthetic data, with models based on synthetic data outperforming those trained on real data in some cases. Furthermore, the resulting models show almost complete immunity to Membership Inference Attacks, manifesting privacy properties missing in models trained with conventional techniques.
Authors: David Debot (Department of Computer Science, KU Leuven), Pietro Barbiero (Universit\`a della Svizzera Italiana,University of Cambridge), Francesco Giannini (Faculty of Sciences, Scuola Normale Superiore, Pisa), Gabriele Ciravegna (Department of Control,Computer Engineering, Politecnico di Torino), Michelangelo Diligenti (Universit\`a di Siena), Giuseppe Marra (Department of Computer Science, KU Leuven)
Abstract: The lack of transparency in the decision-making processes of deep learning systems presents a significant challenge in modern artificial intelligence (AI), as it impairs users' ability to rely on and verify these systems. To address this challenge, Concept Bottleneck Models (CBMs) have made significant progress by incorporating human-interpretable concepts into deep learning architectures. This approach allows predictions to be traced back to specific concept patterns that users can understand and potentially intervene on. However, existing CBMs' task predictors are not fully interpretable, preventing a thorough analysis and any form of formal verification of their decision-making process prior to deployment, thereby raising significant reliability concerns. To bridge this gap, we introduce Concept-based Memory Reasoner (CMR), a novel CBM designed to provide a human-understandable and provably-verifiable task prediction process. Our approach is to model each task prediction as a neural selection mechanism over a memory of learnable logic rules, followed by a symbolic evaluation of the selected rule. The presence of an explicit memory and the symbolic evaluation allow domain experts to inspect and formally verify the validity of certain global properties of interest for the task prediction process. Experimental results demonstrate that CMR achieves comparable accuracy-interpretability trade-offs to state-of-the-art CBMs, discovers logic rules consistent with ground truths, allows for rule interventions, and allows pre-deployment verification.
Authors: Lucas Plagwitz, Lucas Bickmann, Michael Fujarski, Alexander Brenner, Warnes Gobalakrishnan, Lars Eckardt, Antonius B\"uscher, Julian Varghese
Abstract: Electrocardiogram (ECG) recordings have long been vital in diagnosing different cardiac conditions. Recently, research in the field of automatic ECG processing using machine learning methods has gained importance, mainly by utilizing deep learning methods on raw ECG signals. A major advantage of models like convolutional neural networks (CNNs) is their ability to effectively process biomedical imaging or signal data. However, this strength is tempered by challenges related to their lack of explainability, the need for a large amount of training data, and the complexities involved in adapting them for unsupervised clustering tasks. In addressing these tasks, we aim to reintroduce shallow learning techniques, including support vector machines and principal components analysis, into ECG signal processing by leveraging their semi-structured, cyclic form. To this end, we developed and evaluated a transformation that effectively restructures ECG signals into a fully structured format, facilitating their subsequent analysis using shallow learning algorithms. In this study, we present this adaptive transformative approach that aligns R-peaks across all signals in a dataset and resamples the segments between R-peaks, both with and without heart rate dependencies. We illustrate the substantial benefit of this transformation for traditional analysis techniques in the areas of classification, clustering, and explainability, outperforming commercial software for median beat transformation and CNN approaches. Our approach demonstrates a significant advantage for shallow machine learning methods over CNNs, especially when dealing with limited training data. Additionally, we release a fully tested and publicly accessible code framework, providing a robust alignment pipeline to support future research, available at https://github.com/ imi-ms/rlign.
URLs: https://github.com/
Authors: Anthony McNally, Christian Lessig, Peter Lean, Eulalie Boucher, Mihai Alexe, Ewan Pinnington, Matthew Chantry, Simon Lang, Chris Burrows, Marcin Chrust, Florian Pinault, Ethel Villeneuve, Niels Bormann, Sean Healy
Abstract: Skilful Machine Learned weather forecasts have challenged our approach to numerical weather prediction, demonstrating competitive performance compared to traditional physics-based approaches. Data-driven systems have been trained to forecast future weather by learning from long historical records of past weather such as the ECMWF ERA5. These datasets have been made freely available to the wider research community, including the commercial sector, which has been a major factor in the rapid rise of ML forecast systems and the levels of accuracy they have achieved. However, historical reanalyses used for training and real-time analyses used for initial conditions are produced by data assimilation, an optimal blending of observations with a physics-based forecast model. As such, many ML forecast systems have an implicit and unquantified dependence on the physics-based models they seek to challenge. Here we propose a new approach, training a neural network to predict future weather purely from historical observations with no dependence on reanalyses. We use raw observations to initialise a model of the atmosphere (in observation space) learned directly from the observations themselves. Forecasts of crucial weather parameters (such as surface temperature and wind) are obtained by predicting weather parameter observations (e.g. SYNOP surface data) at future times and arbitrary locations. We present preliminary results on forecasting observations 12-hours into the future. These already demonstrate successful learning of time evolutions of the physical processes captured in real observations. We argue that this new approach, by staying purely in observation space, avoids many of the challenges of traditional data assimilation, can exploit a wider range of observations and is readily expanded to simultaneous forecasting of the full Earth system (atmosphere, land, ocean and composition).
Authors: Amir Mohammad Karimi Mamaghan, Samuele Papa, Karl Henrik Johansson, Stefan Bauer, Andrea Dittadi
Abstract: Object-centric (OC) representations, which represent the state of a visual scene by modeling it as a composition of objects, have the potential to be used in various downstream tasks to achieve systematic compositional generalization and facilitate reasoning. However, these claims have not been thoroughly analyzed yet. Recently, foundation models have demonstrated unparalleled capabilities across diverse domains from language to computer vision, marking them as a potential cornerstone of future research for a multitude of computational tasks. In this paper, we conduct an extensive empirical study on representation learning for downstream Visual Question Answering (VQA), which requires an accurate compositional understanding of the scene. We thoroughly investigate the benefits and trade-offs of OC models and alternative approaches including large pre-trained foundation models on both synthetic and real-world data, and demonstrate a viable way to achieve the best of both worlds. The extensiveness of our study, encompassing over 800 downstream VQA models and 15 different types of upstream representations, also provides several additional insights that we believe will be of interest to the community at large.
Authors: Xihong Yang, Yiqi Wang, Jin Chen, Wenqi Fan, Xiangyu Zhao, En Zhu, Xinwang Liu, Defu Lian
Abstract: Deep learning has been widely applied in recommender systems, which has achieved revolutionary progress recently. However, most existing learning-based methods assume that the user and item distributions remain unchanged between the training phase and the test phase. However, the distribution of user and item features can naturally shift in real-world scenarios, potentially resulting in a substantial decrease in recommendation performance. This phenomenon can be formulated as an Out-Of-Distribution (OOD) recommendation problem. To address this challenge, we propose a novel Dual Test-Time-Training framework for OOD Recommendation, termed DT3OR. In DT3OR, we incorporate a model adaptation mechanism during the test-time phase to carefully update the recommendation model, allowing the model to specially adapt to the shifting user and item features. To be specific, we propose a self-distillation task and a contrastive task to assist the model learning both the user's invariant interest preferences and the variant user/item characteristics during the test-time phase, thus facilitating a smooth adaptation to the shifting features. Furthermore, we provide theoretical analysis to support the rationale behind our dual test-time training framework. To the best of our knowledge, this paper is the first work to address OOD recommendation via a test-time-training strategy. We conduct experiments on three datasets with various backbones. Comprehensive experimental results have demonstrated the effectiveness of DT3OR compared to other state-of-the-art baselines.
Authors: Soroosh Tayebi Arasteh, Mahshad Lotfinia, Keno Bressem, Robert Siepmann, Dyke Ferber, Christiane Kuhl, Jakob Nikolas Kather, Sven Nebelung, Daniel Truhn
Abstract: Large language models (LLMs) have advanced the field of artificial intelligence (AI) in medicine. However LLMs often generate outdated or inaccurate information based on static training datasets. Retrieval augmented generation (RAG) mitigates this by integrating outside data sources. While previous RAG systems used pre-assembled, fixed databases with limited flexibility, we have developed Radiology RAG (RadioRAG) as an end-to-end framework that retrieves data from authoritative radiologic online sources in real-time. RadioRAG is evaluated using a dedicated radiologic question-and-answer dataset (RadioQA). We evaluate the diagnostic accuracy of various LLMs when answering radiology-specific questions with and without access to additional online information via RAG. Using 80 questions from RSNA Case Collection across radiologic subspecialties and 24 additional expert-curated questions, for which the correct gold-standard answers were available, LLMs (GPT-3.5-turbo, GPT-4, Mistral-7B, Mixtral-8x7B, and Llama3 [8B and 70B]) were prompted with and without RadioRAG. RadioRAG retrieved context-specific information from www.radiopaedia.org in real-time and incorporated them into its reply. RadioRAG consistently improved diagnostic accuracy across all LLMs, with relative improvements ranging from 2% to 54%. It matched or exceeded question answering without RAG across radiologic subspecialties, particularly in breast imaging and emergency radiology. However, degree of improvement varied among models; GPT-3.5-turbo and Mixtral-8x7B-instruct-v0.1 saw notable gains, while Mistral-7B-instruct-v0.2 showed no improvement, highlighting variability in its effectiveness. LLMs benefit when provided access to domain-specific data beyond their training data. For radiology, RadioRAG establishes a robust framework that substantially improves diagnostic accuracy and factuality in radiological question answering.
Authors: Shahan Nercessian, Johannes Imort, Ninon Devis, Frederik Blang
Abstract: In this paper, we propose and investigate the use of neural audio codec language models for the automatic generation of sample-based musical instruments based on text or reference audio prompts. Our approach extends a generative audio framework to condition on pitch across an 88-key spectrum, velocity, and a combined text/audio embedding. We identify maintaining timbral consistency within the generated instruments as a major challenge. To tackle this issue, we introduce three distinct conditioning schemes. We analyze our methods through objective metrics and human listening tests, demonstrating that our approach can produce compelling musical instruments. Specifically, we introduce a new objective metric to evaluate the timbral consistency of the generated instruments and adapt the average Contrastive Language-Audio Pretraining (CLAP) score for the text-to-instrument case, noting that its naive application is unsuitable for assessing this task. Our findings reveal a complex interplay between timbral consistency, the quality of generated samples, and their correspondence to the input prompt.
Authors: Marco Minici, Federico Cinus, Francesco Bonchi, Giuseppe Manco
Abstract: Signed Graph Neural Networks (SGNNs) have recently gained attention as an effective tool for several learning tasks on signed networks, i.e., graphs where edges have an associated polarity. One of these tasks is to predict the polarity of the links for which this information is missing, starting from the network structure and the other available polarities. However, when the available polarities are few and potentially noisy, such a task becomes challenging. In this work, we devise a semi-supervised learning framework that builds around the novel concept of \emph{multiscale social balance} to improve the prediction of link polarities in settings characterized by limited data quantity and quality. Our model-agnostic approach can seamlessly integrate with any SGNN architecture, dynamically reweighting the importance of each data sample while making strategic use of the structural information from unlabeled edges combined with social balance theory. Empirical validation demonstrates that our approach outperforms established baseline models, effectively addressing the limitations imposed by noisy and sparse data. This result underlines the benefits of incorporating multiscale social balance into SGNNs, opening new avenues for robust and accurate predictions in signed network analysis.
Authors: Claudius Kienle, Benjamin Alt, Onur Celik, Philipp Becker, Darko Katic, Rainer J\"akel, Gerhard Neumann
Abstract: High-level robot skills represent an increasingly popular paradigm in robot programming. However, configuring the skills' parameters for a specific task remains a manual and time-consuming endeavor. Existing approaches for learning or optimizing these parameters often require numerous real-world executions or do not work in dynamic environments. To address these challenges, we propose MuTT, a novel encoder-decoder transformer architecture designed to predict environment-aware executions of robot skills by integrating vision, trajectory, and robot skill parameters. Notably, we pioneer the fusion of vision and trajectory, introducing a novel trajectory projection. Furthermore, we illustrate MuTT's efficacy as a predictor when combined with a model-based robot skill optimizer. This approach facilitates the optimization of robot skill parameters for the current environment, without the need for real-world executions during optimization. Designed for compatibility with any representation of robot skills, MuTT demonstrates its versatility across three comprehensive experiments, showcasing superior performance across two different skill representations.
Authors: Odalric-Ambrym Maillard, Mohammad Sadegh Talebi
Abstract: We consider the situation when a learner faces a set of unknown discrete distributions $(p_k)_{k\in \mathcal K}$ defined over a common alphabet $\mathcal X$, and can build for each distribution $p_k$ an individual high-probability confidence set thanks to $n_k$ observations sampled from $p_k$. The set $(p_k)_{k\in \mathcal K}$ is structured: each distribution $p_k$ is obtained from the same common, but unknown, distribution q via applying an unknown permutation to $\mathcal X$. We call this \emph{permutation-equivalence}. The goal is to build refined confidence sets \emph{exploiting} this structural property. Like other popular notions of structure (Lipschitz smoothness, Linearity, etc.) permutation-equivalence naturally appears in machine learning problems, and to benefit from its potential gain calls for a specific approach. We present a strategy to effectively exploit permutation-equivalence, and provide a finite-time high-probability bound on the size of the refined confidence sets output by the strategy. Since a refinement is not possible for too few observations in general, under mild technical assumptions, our finite-time analysis establish when the number of observations $(n_k)_{k\in \mathcal K}$ are large enough so that the output confidence sets improve over initial individual sets. We carefully characterize this event and the corresponding improvement. Further, our result implies that the size of confidence sets shrink at asymptotic rates of $O(1/\sqrt{\sum_{k\in \mathcal K} n_k})$ and $O(1/\max_{k\in K} n_{k})$, respectively for elements inside and outside the support of q, when the size of each individual confidence set shrinks at respective rates of $O(1/\sqrt{n_k})$ and $O(1/n_k)$. We illustrate the practical benefit of exploiting permutation equivalence on a reinforcement learning task.
Authors: Rasoul Najafi Koopas, Shahed Rezaei, Natalie Rauter, Richard Ostwald, Rolf Lammering
Abstract: A spatiotemporal deep learning framework is proposed that is capable of 2D full-field prediction of fracture in concrete mesostructures. This framework not only predicts fractures but also captures the entire history of the fracture process, from the crack initiation in the interfacial transition zone to the subsequent propagation of the cracks in the mortar matrix. In addition, a convolutional neural network is developed which can predict the averaged stress-strain curve of the mesostructures. The UNet modeling framework, which comprises an encoder-decoder section with skip connections, is used as the deep learning surrogate model. Training and test data are generated from high-fidelity fracture simulations of randomly generated concrete mesostructures. These mesostructures include geometric variabilities such as different aggregate particle geometrical features, spatial distribution, and the total volume fraction of aggregates. The fracture simulations are carried out in Abaqus, utilizing the cohesive phase-field fracture modeling technique as the fracture modeling approach. In this work, to reduce the number of training datasets, the spatial distribution of three sets of material properties for three-phase concrete mesostructures, along with the spatial phase-field damage index, are fed to the UNet to predict the corresponding stress and spatial damage index at the subsequent step. It is shown that after the training process using this methodology, the UNet model is capable of accurately predicting damage on the unseen test dataset by using 470 datasets. Moreover, another novel aspect of this work is the conversion of irregular finite element data into regular grids using a developed pipeline. This approach allows for the implementation of less complex UNet architecture and facilitates the integration of phase-field fracture equations into surrogate models for future developments.
Authors: Zhecan Wang, Garrett Bingham, Adams Yu, Quoc Le, Thang Luong, Golnaz Ghiasi
Abstract: Hallucination has been a major problem for large language models and remains a critical challenge when it comes to multimodality in which vision-language models (VLMs) have to deal with not just textual but also visual inputs. Despite rapid progress in VLMs, resources for evaluating and addressing multimodal hallucination are limited and mostly focused on evaluation. This work introduces HaloQuest, a novel visual question answering dataset that captures various aspects of multimodal hallucination such as false premises, insufficient contexts, and visual challenges. A novel idea from HaloQuest is to leverage synthetic images, apart from real ones, to enable dataset creation at scale. With over 7.7K examples spanning across a wide variety of categories, HaloQuest was designed to be both a challenging benchmark for VLMs and a fine-tuning dataset for advancing multimodal reasoning. Our experiments reveal that current models struggle with HaloQuest, with all open-source VLMs achieving below 36% accuracy. On the other hand, fine-tuning on HaloQuest significantly reduces hallucination rates while preserving performance on standard reasoning tasks. Our results discover that benchmarking with generated images is highly correlated (r=0.97) with real images. Last but not least, we propose a novel Auto-Eval mechanism that is highly correlated with human raters (r=0.99) for evaluating VLMs. In sum, this work makes concrete strides towards understanding, evaluating, and mitigating hallucination in VLMs, serving as an important step towards more reliable multimodal AI systems in the future.
Authors: Daniel Ward, Mark Beaumont, Matteo Fasiolo
Abstract: Estimating a distribution given access to its unnormalized density is pivotal in Bayesian inference, where the posterior is generally known only up to an unknown normalizing constant. Variational inference and Markov chain Monte Carlo methods are the predominant tools for this task; however, both methods are often challenging to apply reliably, particularly when the posterior has complex geometry. Here, we introduce Soft Contrastive Variational Inference (SoftCVI), which allows a family of variational objectives to be derived through a contrastive estimation framework. These objectives have zero variance gradient when the variational approximation is exact, without the need for specialized gradient estimators. The approach involves parameterizing a classifier in terms of the variational distribution, which allows the inference task to be reframed as a contrastive estimation problem, aiming to identify a single true posterior sample among a set of samples. Despite this framing, we do not require positive or negative samples, but rather learn by sampling the variational distribution and computing ground truth soft classification labels from the unnormalized posterior itself. We empirically investigate the performance on a variety of Bayesian inference tasks, using both using both simple (e.g. normal) and expressive (normalizing flow) variational distributions. We find that SoftCVI objectives often outperform other commonly used variational objectives.
Authors: Zhuoyan Xu, Zhenmei Shi, Yingyu Liang
Abstract: Large language models (LLMs) have emerged as powerful tools for many AI problems and exhibit remarkable in-context learning (ICL) capabilities. Compositional ability, solving unseen complex tasks that combine two or more simple tasks, is an essential reasoning ability for Artificial General Intelligence. Despite LLM's tremendous success, how they approach composite tasks, especially those not encountered during the pretraining phase, remains an open question and largely ununderstood. In this study, we delve into the ICL capabilities of LLMs on composite tasks, with only simple tasks as in-context examples. We develop a test suite of composite tasks that include linguistic and logical challenges and perform empirical studies across different LLM families. We observe that models exhibit divergent behaviors: (1) For simpler composite tasks that apply distinct mapping mechanisms to different input segments, the models demonstrate decent compositional ability, while scaling up the model enhances this ability; (2) for more complex composite tasks that involving reasoning multiple steps, where each step represent one task, models typically underperform, and scaling up generally provide no improvements. We offer theoretical analysis in a simplified setting, explaining that models exhibit compositional capability when the task handles different input parts separately. We believe our work sheds new light on the capabilities of LLMs in solving composite tasks regarding the nature of the tasks and model scale. Our dataset and code are available at {\url{https://github.com/OliverXUZY/LLM_Compose}}.
Authors: Josiah Couch, Ramy Arnaout, Rima Arnaout
Abstract: In deep learning, achieving high performance on image classification tasks requires diverse training sets. However, dataset diversity is incompletely understood. The current best practice is to try to maximize dataset size and class balance. Yet large, class-balanced datasets are not guaranteed to be diverse: images can still be arbitrarily similar. We hypothesized that, for a given model architecture, better model performance can be achieved by maximizing dataset diversity more directly. This could open a path for performance improvement without additional computational resources or architectural advances. To test this hypothesis, we introduce a comprehensive framework of diversity measures, developed in ecology, that generalizes familiar quantities like Shannon entropy by accounting for similarities and differences among images. (Dataset size and class balance emerge from this framework as special cases.) By analyzing thousands of subsets from seven medical datasets representing ultrasound, X-ray, CT, and pathology images, we found that the best correlates of performance were not size or class balance but $A$ -- ``big alpha'' -- a set of generalized entropy measures interpreted as the effective number of image-class pairs in the dataset, after accounting for similarities among images. One of these, $A_0$, explained 67\% of the variance in balanced accuracy across all subsets, vs. 54\% for class balance and just 39\% for size. The best pair was size and $A_1$ (79\%), which outperformed size and class balance (74\%). $A$ performed best for subsets from individual datasets as well as across datasets, supporting the generality of these results. We propose maximizing $A$ as a potential new way to improve the performance of deep learning in medical imaging.
Authors: Juan Diego Toscano, Theo K\"aufer, Martin Maxey, Christian Cierpka, George Em Karniadakis
Abstract: We propose the Artificial Intelligence Velocimetry-Thermometry (AIVT) method to infer hidden temperature fields from experimental turbulent velocity data. This physics-informed machine learning method enables us to infer continuous temperature fields using only sparse velocity data, hence eliminating the need for direct temperature measurements. Specifically, AIVT is based on physics-informed Kolmogorov-Arnold Networks (not neural networks) and is trained by optimizing a combined loss function that minimizes the residuals of the velocity data, boundary conditions, and the governing equations. We apply AIVT to a unique set of experimental volumetric and simultaneous temperature and velocity data of Rayleigh-B\'enard convection (RBC) that we acquired by combining Particle Image Thermometry and Lagrangian Particle Tracking. This allows us to compare AIVT predictions and measurements directly. We demonstrate that we can reconstruct and infer continuous and instantaneous velocity and temperature fields from sparse experimental data at a fidelity comparable to direct numerical simulations (DNS) of turbulence. This, in turn, enables us to compute important quantities for quantifying turbulence, such as fluctuations, viscous and thermal dissipation, and QR distribution. This paradigm shift in processing experimental data using AIVT to infer turbulent fields at DNS-level fidelity is a promising avenue in breaking the current deadlock of quantitative understanding of turbulence at high Reynolds numbers, where DNS is computationally infeasible.
Authors: Silvio Galesso, Philipp Schr\"oppel, Hssan Driss, Thomas Brox
Abstract: In recent years, research on out-of-distribution (OoD) detection for semantic segmentation has mainly focused on road scenes -- a domain with a constrained amount of semantic diversity. In this work, we challenge this constraint and extend the domain of this task to general natural images. To this end, we introduce: 1. the ADE-OoD benchmark, which is based on the ADE20k dataset and includes images from diverse domains with a high semantic diversity, and 2. a novel approach that uses Diffusion score matching for OoD detection (DOoD) and is robust to the increased semantic diversity. ADE-OoD features indoor and outdoor images, defines 150 semantic categories as in-distribution, and contains a variety of OoD objects. For DOoD, we train a diffusion model with an MLP architecture on semantic in-distribution embeddings and build on the score matching interpretation to compute pixel-wise OoD scores at inference time. On common road scene OoD benchmarks, DOoD performs on par or better than the state of the art, without using outliers for training or making assumptions about the data domain. On ADE-OoD, DOoD outperforms previous approaches, but leaves much room for future improvements.
Authors: Bunlong Lay, Sebastian Zaczek, Kristina Tesch, Timo Gerkmann
Abstract: Single-channel speech separation is a crucial task for enhancing speech recognition systems in multi-speaker environments. This paper investigates the robustness of state-of-the-art Neural Network models in scenarios where the pitch differences between speakers are minimal. Building on earlier findings by Ditter and Gerkmann, which identified a significant performance drop for the 2018 Chimera++ under similar-pitch conditions, our study extends the analysis to more recent and sophisticated Neural Network models. Our experiments reveal that modern models have substantially reduced the performance gap for matched training and testing conditions. However, a substantial performance gap persists under mismatched conditions, with models performing well for large pitch differences but showing worse performance if the speakers' pitches are similar. These findings motivate further research into the generalizability of speech separation models to similar-pitch speakers and unseen data.
Authors: Haoning Wu, Dongxu Li, Bei Chen, Junnan Li
Abstract: Large multimodal models (LMMs) are processing increasingly longer and richer inputs. Albeit the progress, few public benchmark is available to measure such development. To mitigate this gap, we introduce LongVideoBench, a question-answering benchmark that features video-language interleaved inputs up to an hour long. Our benchmark includes 3,763 varying-length web-collected videos with their subtitles across diverse themes, designed to comprehensively evaluate LMMs on long-term multimodal understanding. To achieve this, we interpret the primary challenge as to accurately retrieve and reason over detailed multimodal information from long inputs. As such, we formulate a novel video question-answering task termed referring reasoning. Specifically, as part of the question, it contains a referring query that references related video contexts, called referred context. The model is then required to reason over relevant video details from the referred context. Following the paradigm of referring reasoning, we curate 6,678 human-annotated multiple-choice questions in 17 fine-grained categories, establishing one of the most comprehensive benchmarks for long-form video understanding. Evaluations suggest that the LongVideoBench presents significant challenges even for the most advanced proprietary models (e.g. GPT-4o, Gemini-1.5-Pro, GPT-4-Turbo), while their open-source counterparts show an even larger performance gap. In addition, our results indicate that model performance on the benchmark improves only when they are capable of processing more frames, positioning LongVideoBench as a valuable benchmark for evaluating future-generation long-context LMMs.
Authors: Artur Miroszewski, Marco Fellous Asiani, Jakub Mielczarek, Bertrand Le Saux, Jakub Nalepa
Abstract: Quantum Machine Learning (QML) has gathered significant attention through approaches like Quantum Kernel Machines. While these methods hold considerable promise, their quantum nature presents inherent challenges. One major challenge is the limited resolution of estimated kernel values caused by the finite number of circuit runs performed on a quantum device. In this study, we propose a comprehensive system of rules and heuristics for estimating the required number of circuit runs in quantum kernel methods. We introduce two critical effects that necessitate an increased measurement precision through additional circuit runs: the spread effect and the concentration effect. The effects are analyzed in the context of fidelity and projected quantum kernels. To address these phenomena, we develop an approach for estimating desired precision of kernel values, which, in turn, is translated into the number of circuit runs. Our methodology is validated through extensive numerical simulations, focusing on the problem of exponential value concentration. We stress that quantum kernel methods should not only be considered from the machine learning performance perspective, but also from the context of the resource consumption. The results provide insights into the possible benefits of quantum kernel methods, offering a guidance for their application in quantum machine learning tasks.
Authors: Emanuele Frascaroli, Aniello Panariello, Pietro Buzzega, Lorenzo Bonicelli, Angelo Porrello, Simone Calderara
Abstract: With the emergence of Transformers and Vision-Language Models (VLMs) such as CLIP, large pre-trained models have become a common strategy to enhance performance in Continual Learning scenarios. This led to the development of numerous prompting strategies to effectively fine-tune transformer-based models without succumbing to catastrophic forgetting. However, these methods struggle to specialize the model on domains significantly deviating from the pre-training and preserving its zero-shot capabilities. In this work, we propose Continual Generative training for Incremental prompt-Learning, a novel approach to mitigate forgetting while adapting a VLM, which exploits generative replay to align prompts to tasks. We also introduce a new metric to evaluate zero-shot capabilities within CL benchmarks. Through extensive experiments on different domains, we demonstrate the effectiveness of our framework in adapting to new tasks while improving zero-shot capabilities. Further analysis reveals that our approach can bridge the gap with joint prompt tuning. The codebase is available at https://github.com/aimagelab/mammoth.
Authors: Vikash Sehwag, Xianghao Kong, Jingtao Li, Michael Spranger, Lingjuan Lyu
Abstract: As scaling laws in generative AI push performance, they also simultaneously concentrate the development of these models among actors with large computational resources. With a focus on text-to-image (T2I) generative models, we aim to address this bottleneck by demonstrating very low-cost training of large-scale T2I diffusion transformer models. As the computational cost of transformers increases with the number of patches in each image, we propose to randomly mask up to 75% of the image patches during training. We propose a deferred masking strategy that preprocesses all patches using a patch-mixer before masking, thus significantly reducing the performance degradation with masking, making it superior to model downscaling in reducing computational cost. We also incorporate the latest improvements in transformer architecture, such as the use of mixture-of-experts layers, to improve performance and further identify the critical benefit of using synthetic images in micro-budget training. Finally, using only 37M publicly available real and synthetic images, we train a 1.16 billion parameter sparse transformer with only \$1,890 economical cost and achieve a 12.7 FID in zero-shot generation on the COCO dataset. Notably, our model achieves competitive FID and high-quality generations while incurring 118$\times$ lower cost than stable diffusion models and 14$\times$ lower cost than the current state-of-the-art approach that costs \$28,400. We aim to release our end-to-end training pipeline to further democratize the training of large-scale diffusion models on micro-budgets.
Authors: Catarina G Belem, Markelle Kelly, Mark Steyvers, Sameer Singh, Padhraic Smyth
Abstract: Uncertainty expressions such as ``probably'' or ``highly unlikely'' are pervasive in human language. While prior work has established that there is population-level agreement in terms of how humans interpret these expressions, there has been little inquiry into the abilities of language models to interpret such expressions. In this paper, we investigate how language models map linguistic expressions of uncertainty to numerical responses. Our approach assesses whether language models can employ theory of mind in this setting: understanding the uncertainty of another agent about a particular statement, independently of the model's own certainty about that statement. We evaluate both humans and 10 popular language models on a task created to assess these abilities. Unexpectedly, we find that 8 out of 10 models are able to map uncertainty expressions to probabilistic responses in a human-like manner. However, we observe systematically different behavior depending on whether a statement is actually true or false. This sensitivity indicates that language models are substantially more susceptible to bias based on their prior knowledge (as compared to humans). These findings raise important questions and have broad implications for human-AI alignment and AI-AI communication.
Authors: Eugene Valassakis, Guillermo Garcia-Hernando
Abstract: Predicting camera-space hand meshes from single RGB images is crucial for enabling realistic hand interactions in 3D virtual and augmented worlds. Previous work typically divided the task into two stages: given a cropped image of the hand, predict meshes in relative coordinates, followed by lifting these predictions into camera space in a separate and independent stage, often resulting in the loss of valuable contextual and scale information. To prevent the loss of these cues, we propose unifying these two stages into an end-to-end solution that addresses the 2D-3D correspondence problem. This solution enables back-propagation from camera space outputs to the rest of the network through a new differentiable global positioning module. We also introduce an image rectification step that harmonizes both the training dataset and the input image as if they were acquired with the same camera, helping to alleviate the inherent scale-depth ambiguity of the problem. We validate the effectiveness of our framework in evaluations against several baselines and state-of-the-art approaches across three public benchmarks.
Authors: Bernd Fritzke (individual researcher)
Abstract: We introduce the breathing k-means algorithm, which significantly improves upon the widely-known greedy k-means++ algorithm, the default method for k-means clustering in the scikit-learn package. Our approach is able to improve solutions obtained by greedy k-means++ through a novel 'breathing' technique cyclically increasing and decreasing the number of centroids based on local error and utility measures. We conducted experiments using greedy k-means++ as a baseline, comparing it with breathing k-means and five other k-means algorithms. Among the methods investigated, only breathing k-means and better k-means++ consistently outperformed the baseline, with breathing k-means demonstrating a substantial lead. This superior performance was maintained even when comparing the best result of ten runs for all other algorithms to a single run of breathing k-means, demonstrating its effectiveness and speed. Our findings indicate that the breathing k-means algorithm outperforms the other k-means techniques, especially greedy k-means++ with ten repetitions, which it dominates in both solution quality and speed. This positions breathing k-means as a full replacement for greedy k-means++.
Authors: Hongseok Namkoong, Samuel Daulton, Eytan Bakshy
Abstract: Thompson sampling (TS) has emerged as a robust technique for contextual bandit problems. However, TS requires posterior inference and optimization for action generation, prohibiting its use in many online platforms where latency and ease of deployment are of concern. We operationalize TS by proposing a novel imitation-learning-based algorithm that distills a TS policy into an explicit policy representation, allowing fast decision-making and easy deployment in mobile and server-based environments. Using batched data collected under the imitation policy, our algorithm iteratively performs offline updates to the TS policy, and learns a new explicit policy representation to imitate it. Empirically, our imitation policy achieves performance comparable to batch TS while allowing more than an order of magnitude reduction in decision-time latency. Buoyed by low latency and simplicity of implementation, our algorithm has been successfully deployed in multiple video upload systems for Meta. Using a randomized controlled trial, we show our algorithm resulted in significant improvements in video quality and watch time.
Authors: Jing-Cheng Pang, Tian Xu, Shengyi Jiang, Yu-Ren Liu, Yang Yu
Abstract: Reinforcement learning (RL) has demonstrated impressive performance in decision-making tasks like embodied control, autonomous driving and financial trading. In many decision-making tasks, the agents often encounter the problem of executing actions under limited budgets. However, classic RL methods typically overlook the challenges posed by such sparse-executing actions. They operate under the assumption that all actions can be taken for a unlimited number of times, both in the formulation of the problem and in the development of effective algorithms. To tackle the issue of limited action execution in RL, this paper first formalizes the problem as a Sparse Action Markov Decision Process (SA-MDP), in which specific actions in the action space can only be executed for a limited time. Then, we propose a policy optimization algorithm, Action Sparsity REgularization (ASRE), which adaptively handles each action with a distinct preference. ASRE operates through two steps: First, ASRE evaluates action sparsity by constrained action sampling. Following this, ASRE incorporates the sparsity evaluation into policy learning by way of an action distribution regularization. We provide theoretical identification that validates the convergence of ASRE to a regularized optimal value function. Experiments on tasks with known sparse-executing actions, where classical RL algorithms struggle to train policy efficiently, ASRE effectively constrains the action sampling and outperforms baselines. Moreover, we present that ASRE can generally improve the performance in Atari games, demonstrating its broad applicability.
Authors: Mark Lowell
Abstract: Despite the widespread adoption of neural networks, their training dynamics remain poorly understood. We show experimentally that as the size of the dataset increases, a point forms where the magnitude of the gradient of the loss becomes unbounded. Gradient descent rapidly brings the network close to this singularity in parameter space, and further training takes place near it. This singularity explains a variety of phenomena recently observed in the Hessian of neural network loss functions, such as training on the edge of stability and the concentration of the gradient in a top subspace. Once the network approaches the singularity, the top subspace contributes little to learning, even though it constitutes the majority of the gradient.
Authors: A N M Nafiz Abeer, Nathan Urban, M Ryan Weil, Francis J. Alexander, Byung-Jun Yoon
Abstract: Molecular design based on generative models, such as variational autoencoders (VAEs), has become increasingly popular in recent years due to its efficiency for exploring high-dimensional molecular space to identify molecules with desired properties. While the efficacy of the initial model strongly depends on the training data, the sampling efficiency of the model for suggesting novel molecules with enhanced properties can be further enhanced via latent space optimization. In this paper, we propose a multi-objective latent space optimization (LSO) method that can significantly enhance the performance of generative molecular design (GMD). The proposed method adopts an iterative weighted retraining approach, where the respective weights of the molecules in the training data are determined by their Pareto efficiency. We demonstrate that our multi-objective GMD LSO method can significantly improve the performance of GMD for jointly optimizing multiple molecular properties.
Authors: Florinel-Alin Croitoru, Nicolae-Catalin Ristea, Radu Tudor Ionescu, Nicu Sebe
Abstract: Most curriculum learning methods require an approach to sort the data samples by difficulty, which is often cumbersome to perform. In this work, we propose a novel curriculum learning approach termed Learning Rate Curriculum (LeRaC), which leverages the use of a different learning rate for each layer of a neural network to create a data-agnostic curriculum during the initial training epochs. More specifically, LeRaC assigns higher learning rates to neural layers closer to the input, gradually decreasing the learning rates as the layers are placed farther away from the input. The learning rates increase at various paces during the first training iterations, until they all reach the same value. From this point on, the neural model is trained as usual. This creates a model-level curriculum learning strategy that does not require sorting the examples by difficulty and is compatible with any neural network, generating higher performance levels regardless of the architecture. We conduct comprehensive experiments on 12 data sets from the computer vision (CIFAR-10, CIFAR-100, Tiny ImageNet, ImageNet-200, Food-101, UTKFace, PASCAL VOC), language (BoolQ, QNLI, RTE) and audio (ESC-50, CREMA-D) domains, considering various convolutional (ResNet-18, Wide-ResNet-50, DenseNet-121, YOLOv5), recurrent (LSTM) and transformer (CvT, BERT, SepTr) architectures. We compare our approach with the conventional training regime, as well as with Curriculum by Smoothing (CBS), a state-of-the-art data-agnostic curriculum learning approach. Unlike CBS, our performance improvements over the standard training regime are consistent across all data sets and models. Furthermore, we significantly surpass CBS in terms of training time (there is no additional cost over the standard training regime for LeRaC). Our code is freely available at: https://github.com/CroitoruAlin/LeRaC.
Authors: Chengli Tan, Jiangshe Zhang, Junmin Liu
Abstract: One of the fundamental challenges in the deep learning community is to theoretically understand how well a deep neural network generalizes to unseen data. However, current approaches often yield generalization bounds that are either too loose to be informative of the true generalization error or only valid to the compressed nets. In this study, we present a simple yet non-vacuous generalization bound from the optimization perspective. We achieve this goal by leveraging that the hypothesis set accessed by stochastic gradient algorithms is essentially fractal-like and thus can derive a tighter bound over the algorithm-dependent Rademacher complexity. The main argument rests on modeling the discrete-time recursion process via a continuous-time stochastic differential equation driven by fractional Brownian motion. Numerical studies demonstrate that our approach is able to yield plausible generalization guarantees for modern neural networks such as ResNet and Vision Transformer, even when they are trained on a large-scale dataset (e.g. ImageNet-1K).
Authors: Banafsheh Rafiee, Sina Ghiassian, Jun Jin, Richard Sutton, Jun Luo, Adam White
Abstract: In this paper, we explore an approach to auxiliary task discovery in reinforcement learning based on ideas from representation learning. Auxiliary tasks tend to improve data efficiency by forcing the agent to learn auxiliary prediction and control objectives in addition to the main task of maximizing reward, and thus producing better representations. Typically these tasks are designed by people. Meta-learning offers a promising avenue for automatic task discovery; however, these methods are computationally expensive and challenging to tune in practice. In this paper, we explore a complementary approach to the auxiliary task discovery: continually generating new auxiliary tasks and preserving only those with high utility. We also introduce a new measure of auxiliary tasks' usefulness based on how useful the features induced by them are for the main task. Our discovery algorithm significantly outperforms random tasks and learning without auxiliary tasks across a suite of environments.
Authors: Rui Zhu, Di Tang, Siyuan Tang, XiaoFeng Wang, Haixu Tang
Abstract: In this paper, we present a simple yet surprisingly effective technique to induce "selective amnesia" on a backdoored model. Our approach, called SEAM, has been inspired by the problem of catastrophic forgetting (CF), a long standing issue in continual learning. Our idea is to retrain a given DNN model on randomly labeled clean data, to induce a CF on the model, leading to a sudden forget on both primary and backdoor tasks; then we recover the primary task by retraining the randomized model on correctly labeled clean data. We analyzed SEAM by modeling the unlearning process as continual learning and further approximating a DNN using Neural Tangent Kernel for measuring CF. Our analysis shows that our random-labeling approach actually maximizes the CF on an unknown backdoor in the absence of triggered inputs, and also preserves some feature extraction in the network to enable a fast revival of the primary task. We further evaluated SEAM on both image processing and Natural Language Processing tasks, under both data contamination and training manipulation attacks, over thousands of models either trained on popular image datasets or provided by the TrojAI competition. Our experiments show that SEAM vastly outperforms the state-of-the-art unlearning techniques, achieving a high Fidelity (measuring the gap between the accuracy of the primary task and that of the backdoor) within a few minutes (about 30 times faster than training a model from scratch using the MNIST dataset), with only a small amount of clean data (0.1% of training data for TrojAI models).
Authors: Xiangfeng Wang, Hongteng Xu, Moyi Yang
Abstract: Distributed distribution comparison aims to measure the distance between the distributions whose data are scattered across different agents in a distributed system and cannot even be shared directly among the agents. In this study, we propose a novel decentralized entropic optimal transport (DEOT) method, which provides a communication-efficient and privacy-preserving solution to this problem with theoretical guarantees. In particular, we design a mini-batch randomized block-coordinate descent (MRBCD) scheme to optimize the DEOT distance in its dual form. The dual variables are scattered across different agents and updated locally and iteratively with limited communications among partial agents. The kernel matrix involved in the gradients of the dual variables is estimated by a decentralized kernel approximation method, in which each agent only needs to approximate and store a sub-kernel matrix by one-shot communication and without sharing raw data. Besides computing entropic Wasserstein distance, we show that the proposed MRBCD scheme and kernel approximation method also apply to entropic Gromov-Wasserstein distance. We analyze our method's communication complexity and, under mild assumptions, provide a theoretical bound for the approximation error caused by the convergence error, the estimated kernel, and the mismatch between the storage and communication protocols. In addition, we discuss the trade-off between the precision of the EOT distance and the strength of privacy protection when implementing our method. Experiments on synthetic data and real-world distributed domain adaptation tasks demonstrate the effectiveness of our method.
Authors: Haozhe Feng, Tianyu Pang, Chao Du, Wei Chen, Shuicheng Yan, Min Lin
Abstract: Federated learning (FL) is a general principle for decentralized clients to train a server model collectively without sharing local data. FL is a promising framework with practical applications, but its standard training paradigm requires the clients to backpropagate through the model to compute gradients. Since these clients are typically edge devices and not fully trusted, executing backpropagation on them incurs computational and storage overhead as well as white-box vulnerability. In light of this, we develop backpropagation-free federated learning, dubbed BAFFLE, in which backpropagation is replaced by multiple forward processes to estimate gradients. BAFFLE is 1) memory-efficient and easily fits uploading bandwidth; 2) compatible with inference-only hardware optimization and model quantization or pruning; and 3) well-suited to trusted execution environments, because the clients in BAFFLE only execute forward propagation and return a set of scalars to the server. Empirically we use BAFFLE to train deep models from scratch or to finetune pretrained models, achieving acceptable results. Code is available in https://github.com/FengHZ/BAFFLE.
Authors: Yatong Bai, Brendon G. Anderson, Aerin Kim, Somayeh Sojoudi
Abstract: While prior research has proposed a plethora of methods that build neural classifiers robust against adversarial robustness, practitioners are still reluctant to adopt them due to their unacceptably severe clean accuracy penalties. This paper significantly alleviates this accuracy-robustness trade-off by mixing the output probabilities of a standard classifier and a robust classifier, where the standard network is optimized for clean accuracy and is not robust in general. We show that the robust base classifier's confidence difference for correct and incorrect examples is the key to this improvement. In addition to providing intuitions and empirical evidence, we theoretically certify the robustness of the mixed classifier under realistic assumptions. Furthermore, we adapt an adversarial input detector into a mixing network that adaptively adjusts the mixture of the two base models, further reducing the accuracy penalty of achieving robustness. The proposed flexible method, termed "adaptive smoothing", can work in conjunction with existing or even future methods that improve clean accuracy, robustness, or adversary detection. Our empirical evaluation considers strong attack methods, including AutoAttack and adaptive attack. On the CIFAR-100 dataset, our method achieves an 85.21% clean accuracy while maintaining a 38.72% $\ell_\infty$-AutoAttacked ($\epsilon = 8/255$) accuracy, becoming the second most robust method on the RobustBench CIFAR-100 benchmark as of submission, while improving the clean accuracy by ten percentage points compared with all listed models. The code that implements our method is available at https://github.com/Bai-YT/AdaptiveSmoothing.
Authors: Daniel McKenzie, Samy Wu Fung, Howard Heaton
Abstract: In many applications, a combinatorial problem must be repeatedly solved with similar, but distinct parameters. Yet, the parameters $w$ are not directly observed; only contextual data $d$ that correlates with $w$ is available. It is tempting to use a neural network to predict $w$ given $d$. However, training such a model requires reconciling the discrete nature of combinatorial optimization with the gradient-based frameworks used to train neural networks. We study the case where the problem in question is an Integer Linear Program (ILP). We propose applying a three-operator splitting technique, also known as Davis-Yin splitting (DYS), to the quadratically regularized continuous relaxation of the ILP. We prove that the resulting scheme is compatible with the recently introduced Jacobian-free backpropagation (JFB). Our experiments on two representative ILPs: the shortest path problem and the knapsack problem, demonstrate that this combination-DYS on the forward pass, JFB on the backward pass-yields a scheme which scales more effectively to high-dimensional problems than existing schemes. All code associated with this paper is available at github.com/mines-opt-ml/fpo-dys.
Authors: Yue Wu, Yewen Fan, Paul Pu Liang, Amos Azaria, Yuanzhi Li, Tom M. Mitchell
Abstract: High sample complexity has long been a challenge for RL. On the other hand, humans learn to perform tasks not only from interaction or demonstrations, but also by reading unstructured text documents, e.g., instruction manuals. Instruction manuals and wiki pages are among the most abundant data that could inform agents of valuable features and policies or task-specific environmental dynamics and reward structures. Therefore, we hypothesize that the ability to utilize human-written instruction manuals to assist learning policies for specific tasks should lead to a more efficient and better-performing agent. We propose the Read and Reward framework. Read and Reward speeds up RL algorithms on Atari games by reading manuals released by the Atari game developers. Our framework consists of a QA Extraction module that extracts and summarizes relevant information from the manual and a Reasoning module that evaluates object-agent interactions based on information from the manual. An auxiliary reward is then provided to a standard A2C RL agent, when interaction is detected. Experimentally, various RL algorithms obtain significant improvement in performance and training speed when assisted by our design.
Authors: Sam van der Poel, Dakotah Lambert, Kalina Kostyszyn, Tiantian Gao, Rahul Verma, Derek Andersen, Joanne Chau, Emily Peterson, Cody St. Clair, Paul Fodor, Chihiro Shibata, Jeffrey Heinz
Abstract: Synthetic datasets constructed from formal languages allow fine-grained examination of the learning and generalization capabilities of machine learning systems for sequence classification. This article presents a new benchmark for machine learning systems on sequence classification called MLRegTest, which contains training, development, and test sets from 1,800 regular languages. Different kinds of formal languages represent different kinds of long-distance dependencies, and correctly identifying long-distance dependencies in sequences is a known challenge for ML systems to generalize successfully. MLRegTest organizes its languages according to their logical complexity (monadic second order, first order, propositional, or monomial expressions) and the kind of logical literals (string, tier-string, subsequence, or combinations thereof). The logical complexity and choice of literal provides a systematic way to understand different kinds of long-distance dependencies in regular languages, and therefore to understand the capacities of different ML systems to learn such long-distance dependencies. Finally, the performance of different neural networks (simple RNN, LSTM, GRU, transformer) on MLRegTest is examined. The main conclusion is that performance depends significantly on the kind of test set, the class of language, and the neural network architecture.
Authors: Zheng Zhai, Hengchao Chen, Zhigang Yao
Abstract: Ridge estimation is an important manifold learning technique. The goal of this paper is to examine the effects of nonlinear transformations on the ridge sets. The main result proves the inclusion relationship between ridges: $\cR(f\circ p)\subseteq \cR(p)$, provided that the transformation $f$ is strictly increasing and concave on the range of the function $p$. Additionally, given an underlying true manifold $\cM$, we show that the Hausdorff distance between $\cR(f\circ p)$ and its projection onto $\cM$ is smaller than the Hausdorff distance between $\cR(p)$ and the corresponding projection. This motivates us to apply an increasing and concave transformation before the ridge estimation. In specific, we show that the power transformations $f^{q}(y)=y^q/q,-\infty
Authors: Kai Zhao, Jianye Hao, Yi Ma, Jinyi Liu, Yan Zheng, Zhaopeng Meng
Abstract: Offline reinforcement learning (RL) is a learning paradigm where an agent learns from a fixed dataset of experience. However, learning solely from a static dataset can limit the performance due to the lack of exploration. To overcome it, offline-to-online RL combines offline pre-training with online fine-tuning, which enables the agent to further refine its policy by interacting with the environment in real-time. Despite its benefits, existing offline-to-online RL methods suffer from performance degradation and slow improvement during the online phase. To tackle these challenges, we propose a novel framework called ENsemble-based Offline-To-Online (ENOTO) RL. By increasing the number of Q-networks, we seamlessly bridge offline pre-training and online fine-tuning without degrading performance. Moreover, to expedite online performance enhancement, we appropriately loosen the pessimism of Q-value estimation and incorporate ensemble-based exploration mechanisms into our framework. Experimental results demonstrate that ENOTO can substantially improve the training stability, learning efficiency, and final performance of existing offline RL methods during online fine-tuning on a range of locomotion and navigation tasks, significantly outperforming existing offline-to-online RL methods.
Authors: Jiapeng Wu, Atiyeh Ashari Ghomi, David Glukhov, Jesse C. Cresswell, Franziska Boenisch, Nicolas Papernot
Abstract: Machine learning models are susceptible to a variety of attacks that can erode trust, including attacks against the privacy of training data, and adversarial examples that jeopardize model accuracy. Differential privacy and certified robustness are effective frameworks for combating these two threats respectively, as they each provide future-proof guarantees. However, we show that standard differentially private model training is insufficient for providing strong certified robustness guarantees. Indeed, combining differential privacy and certified robustness in a single system is non-trivial, leading previous works to introduce complex training schemes that lack flexibility. In this work, we present DP-CERT, a simple and effective method that achieves both privacy and robustness guarantees simultaneously by integrating randomized smoothing into standard differentially private model training. Compared to the leading prior work, DP-CERT gives up to a 2.5% increase in certified accuracy for the same differential privacy guarantee on CIFAR10. Through in-depth persample metric analysis, we find that larger certifiable radii correlate with smaller local Lipschitz constants, and show that DP-CERT effectively reduces Lipschitz constants compared to other differentially private training methods. The code is available at github.com/layer6ailabs/dp-cert.
Authors: Yang Xu, Yunming Liao, Hongli Xu, Zhipeng Sun, Liusheng Huang, Chunming Qiao
Abstract: Federated Learning (FL) has emerged to allow multiple clients to collaboratively train machine learning models on their private data at the network edge. However, training and deploying large-scale models on resource-constrained devices is challenging. Fortunately, Split Federated Learning (SFL) offers a feasible solution by alleviating the computation and/or communication burden on clients. However, existing SFL works often assume sufficient labeled data on clients, which is usually impractical. Besides, data non-IIDness poses another challenge to ensure efficient model training. To our best knowledge, the above two issues have not been simultaneously addressed in SFL. Herein, we propose a novel Semi-supervised SFL system, termed SemiSFL, which incorporates clustering regularization to perform SFL with unlabeled and non-IID client data. Moreover, our theoretical and experimental investigations into model convergence reveal that the inconsistent training processes on labeled and unlabeled data have an influence on the effectiveness of clustering regularization. To mitigate the training inconsistency, we develop an algorithm for dynamically adjusting the global updating frequency, so as to improve training performance. Extensive experiments on benchmark models and datasets show that our system provides a 3.8x speed-up in training time, reduces the communication cost by about 70.3% while reaching the target accuracy, and achieves up to 5.8% improvement in accuracy under non-IID scenarios compared to the state-of-the-art baselines.
Authors: Parvin Malekzadeh, Ming Hou, Konstantinos N. Plataniotis
Abstract: Sample efficiency is central to developing practical reinforcement learning (RL) for complex and large-scale decision-making problems. The ability to transfer and generalize knowledge gained from previous experiences to downstream tasks can significantly improve sample efficiency. Recent research indicates that successor feature (SF) RL algorithms enable knowledge generalization between tasks with different rewards but identical transition dynamics. It has recently been hypothesized that combining model-based (MB) methods with SF algorithms can alleviate the limitation of fixed transition dynamics. Furthermore, uncertainty-aware exploration is widely recognized as another appealing approach for improving sample efficiency. Putting together two ideas of hybrid model-based successor feature (MB-SF) and uncertainty leads to an approach to the problem of sample efficient uncertainty-aware knowledge transfer across tasks with different transition dynamics or/and reward functions. In this paper, the uncertainty of the value of each action is approximated by a Kalman filter (KF)-based multiple-model adaptive estimation. This KF-based framework treats the parameters of a model as random variables. To the best of our knowledge, this is the first attempt at formulating a hybrid MB-SF algorithm capable of generalizing knowledge across large or continuous state space tasks with various transition dynamics while requiring less computation at decision time than MB methods. The number of samples required to learn the tasks was compared to recent SF and MB baselines. The results show that our algorithm generalizes its knowledge across different transition dynamics, learns downstream tasks with significantly fewer samples than starting from scratch, and outperforms existing approaches.
Authors: Yinan Li, Chicheng Zhang
Abstract: We study the problem of computationally and label efficient PAC active learning $d$-dimensional halfspaces with Tsybakov Noise~\citep{tsybakov2004optimal} under structured unlabeled data distributions. Inspired by~\cite{diakonikolas2020learning}, we prove that any approximate first-order stationary point of a smooth nonconvex loss function yields a halfspace with a low excess error guarantee. In light of the above structural result, we design a nonconvex optimization-based algorithm with a label complexity of $\tilde{O}(d (\frac{1}{\epsilon})^{\frac{8-6\alpha}{3\alpha-1}})$, under the assumption that the Tsybakov noise parameter $\alpha \in (\frac13, 1]$, which narrows down the gap between the label complexities of the previously known efficient passive or active algorithms~\citep{diakonikolas2020polynomial,zhang2021improved} and the information-theoretic lower bound in this setting.
Authors: Mai Zeng, Florence Regol, Mark Coates
Abstract: Neural Temporal Point Processes (TPPs) have emerged as the primary framework for predicting sequences of events that occur at irregular time intervals, but their sequential nature can hamper performance for long-horizon forecasts. To address this, we introduce a novel approach that incorporates a diffusion generative model. The model facilitates sequence-to-sequence prediction, allowing multi-step predictions based on historical event sequences. In contrast to previous approaches, our model directly learns the joint probability distribution of types and inter-arrival times for multiple events. This allows us to fully leverage the high dimensional modeling capability of modern generative models. Our model is composed of two diffusion processes, one for the time intervals and one for the event types. These processes interact through their respective denoising functions, which can take as input intermediate representations from both processes, allowing the model to learn complex interactions. We demonstrate that our proposal outperforms state-of-the-art baselines for long-horizon forecasting of TPP.
Authors: Junu Kim, Chaeeun Shim, Bosco Seong Kyu Yang, Chami Im, Sung Yoon Lim, Han-Gil Jeong, Edward Choi
Abstract: Machine learning (ML) has recently shown promising results in medical predictions using electronic health records (EHRs). However, since ML models typically have a limited capability in terms of input sizes, selecting specific medical events from EHRs for use as input is necessary. This selection process, often relying on expert opinion, can cause bottlenecks in development. We propose Retrieval-Enhanced Medical prediction model (REMed) to address such challenges. REMed can essentially evaluate unlimited medical events, select the relevant ones, and make predictions. This allows for an unrestricted input size, eliminating the need for manual event selection. We verified these properties through experiments involving 27 clinical prediction tasks across four independent cohorts, where REMed outperformed the baselines. Notably, we found that the preferences of REMed align closely with those of medical experts. We expect our approach to significantly expedite the development of EHR prediction models by minimizing clinicians' need for manual involvement.
Authors: Hanlin Zhang, Benjamin L. Edelman, Danilo Francati, Daniele Venturi, Giuseppe Ateniese, Boaz Barak
Abstract: Watermarking generative models consists of planting a statistical signal (watermark) in a model's output so that it can be later verified that the output was generated by the given model. A strong watermarking scheme satisfies the property that a computationally bounded attacker cannot erase the watermark without causing significant quality degradation. In this paper, we study the (im)possibility of strong watermarking schemes. We prove that, under well-specified and natural assumptions, strong watermarking is impossible to achieve. This holds even in the private detection algorithm setting, where the watermark insertion and detection algorithms share a secret key, unknown to the attacker. To prove this result, we introduce a generic efficient watermark attack; the attacker is not required to know the private key of the scheme or even which scheme is used. Our attack is based on two assumptions: (1) The attacker has access to a "quality oracle" that can evaluate whether a candidate output is a high-quality response to a prompt, and (2) The attacker has access to a "perturbation oracle" which can modify an output with a nontrivial probability of maintaining quality, and which induces an efficiently mixing random walk on high-quality outputs. We argue that both assumptions can be satisfied in practice by an attacker with weaker computational capabilities than the watermarked model itself, to which the attacker has only black-box access. Furthermore, our assumptions will likely only be easier to satisfy over time as models grow in capabilities and modalities. We demonstrate the feasibility of our attack by instantiating it to attack three existing watermarking schemes for large language models: Kirchenbauer et al. (2023), Kuditipudi et al. (2023), and Zhao et al. (2023). The same attack successfully removes the watermarks planted by all three schemes, with only minor quality degradation.
Authors: Doyoung Kim, Susik Yoon, Dongmin Park, Youngjun Lee, Hwanjun Song, Jihwan Bang, Jae-Gil Lee
Abstract: In real-world continual learning (CL) scenarios, tasks often exhibit intricate and unpredictable semantic shifts, posing challenges for fixed prompt management strategies which are tailored to only handle semantic shifts of uniform degree (i.e., uniformly mild or uniformly abrupt). To address this limitation, we propose an adaptive prompting approach that effectively accommodates semantic shifts of varying degree where mild and abrupt shifts are mixed. AdaPromptCL employs the assign-and-refine semantic grouping mechanism that dynamically manages prompt groups in accordance with the semantic similarity between tasks, enhancing the quality of grouping through continuous refinement. Our experiment results demonstrate that AdaPromptCL outperforms existing prompting methods by up to 21.3%, especially in the benchmark datasets with diverse semantic shifts between tasks.
Authors: Yunming Liao, Yang Xu, Hongli Xu, Lun Wang, Zhiwei Yao, Chunming Qiao
Abstract: Recently, federated learning (FL) has emerged as a popular technique for edge AI to mine valuable knowledge in edge computing (EC) systems. To mitigate the computing/communication burden on resource-constrained workers and protect model privacy, split federated learning (SFL) has been released by integrating both data and model parallelism. Despite resource limitations, SFL still faces two other critical challenges in EC, i.e., statistical heterogeneity and system heterogeneity. To address these challenges, we propose a novel SFL framework, termed MergeSFL, by incorporating feature merging and batch size regulation in SFL. Concretely, feature merging aims to merge the features from workers into a mixed feature sequence, which is approximately equivalent to the features derived from IID data and is employed to promote model accuracy. While batch size regulation aims to assign diverse and suitable batch sizes for heterogeneous workers to improve training efficiency. Moreover, MergeSFL explores to jointly optimize these two strategies upon their coupled relationship to better enhance the performance of SFL. Extensive experiments are conducted on a physical platform with 80 NVIDIA Jetson edge devices, and the experimental results show that MergeSFL can improve the final model accuracy by 5.82% to 26.22%, with a speedup by about 1.74x to 4.14x, compared to the baselines.
Authors: Zhidi Lin, Yiyong Sun, Feng Yin, Alexandre Hoang Thi\'ery
Abstract: The Gaussian process state-space models (GPSSMs) represent a versatile class of data-driven nonlinear dynamical system models. However, the presence of numerous latent variables in GPSSM incurs unresolved issues for existing variational inference approaches, particularly under the more realistic non-mean-field (NMF) assumption, including extensive training effort, compromised inference accuracy, and infeasibility for online applications, among others. In this paper, we tackle these challenges by incorporating the ensemble Kalman filter (EnKF), a well-established model-based filtering technique, into the NMF variational inference framework to approximate the posterior distribution of the latent states. This novel marriage between EnKF and GPSSM not only eliminates the need for extensive parameterization in learning variational distributions, but also enables an interpretable, closed-form approximation of the evidence lower bound (ELBO). Moreover, owing to the streamlined parameterization via the EnKF, the new GPSSM model can be easily accommodated in online learning applications. We demonstrate that the resulting EnKF-aided online algorithm embodies a principled objective function by ensuring data-fitting accuracy while incorporating model regularizations to mitigate overfitting. We also provide detailed analysis and fresh insights for the proposed algorithms. Comprehensive evaluation across diverse real and synthetic datasets corroborates the superior learning and inference performance of our EnKF-aided variational inference algorithms compared to existing methods.
Authors: M\'elanie Ducoffe, Guillaume Pov\'eda, Audrey Galametz, Ryma Boumazouza, Marion-C\'ecile Martin, Julien Baris, Derk Daverschot, Eugene O'Higgins
Abstract: Surrogate Neural Networks are nowadays routinely used in industry as substitutes for computationally demanding engineering simulations (e.g., in structural analysis). They allow to generate faster predictions and thus analyses in industrial applications e.g., during a product design, testing or monitoring phases. Due to their performance and time-efficiency, these surrogate models are now being developed for use in safety-critical applications. Neural network verification and in particular the assessment of their robustness (e.g., to perturbations) is the next critical step to allow their inclusion in real-life applications and certification. We assess the applicability and scalability of empirical and formal methods in the context of aircraft predictive maintenance for surrogate neural networks designed to predict the stress sustained by an aircraft part from external loads. The case study covers a high-dimensional input and output space and the verification process thus accommodates multi-objective constraints. We explore the complementarity of verification methods in assessing the local stability property of such surrogate models to input noise. We showcase the effectiveness of sequentially combining methods in one verification 'pipeline' and demonstrate the subsequent gain in runtime required to assess the targeted property.
Authors: Xinyi Gao, Junliang Yu, Tong Chen, Guanhua Ye, Wentao Zhang, Hongzhi Yin
Abstract: The rapid growth of graph data poses significant challenges in storage, transmission, and particularly the training of graph neural networks (GNNs). To address these challenges, graph condensation (GC) has emerged as an innovative solution. GC focuses on synthesizing a compact yet highly representative graph, enabling GNNs trained on it to achieve performance comparable to those trained on the original large graph. The notable efficacy of GC and its broad prospects have garnered significant attention and spurred extensive research. This survey paper provides an up-to-date and systematic overview of GC, organizing existing research into five categories aligned with critical GC evaluation criteria: effectiveness, generalization, efficiency, fairness, and robustness. To facilitate an in-depth and comprehensive understanding of GC, this paper examines various methods under each category and thoroughly discusses two essential components within GC: optimization strategies and condensed graph generation. We also empirically compare and analyze representative GC methods with diverse optimization strategies based on the five proposed GC evaluation criteria. Finally, we explore the applications of GC in various fields, outline the related open-source libraries, and highlight the present challenges and novel insights, with the aim of promoting advancements in future research. The related resources can be found at https://github.com/XYGaoG/Graph-Condensation-Papers.
Authors: Hugo Thimonier, Fabrice Popineau, Arpad Rimmel, Bich-Li\^en Doan
Abstract: Deep learning for tabular data has garnered increasing attention in recent years, yet employing deep models for structured data remains challenging. While these models excel with unstructured data, their efficacy with structured data has been limited. Recent research has introduced retrieval-augmented models to address this gap, demonstrating promising results in supervised tasks such as classification and regression. In this work, we investigate using retrieval-augmented models for anomaly detection on tabular data. We propose a reconstruction-based approach in which a transformer model learns to reconstruct masked features of \textit{normal} samples. We test the effectiveness of KNN-based and attention-based modules to select relevant samples to help in the reconstruction process of the target sample. Our experiments on a benchmark of 31 tabular datasets reveal that augmenting this reconstruction-based anomaly detection (AD) method with sample-sample dependencies via retrieval modules significantly boosts performance. The present work supports the idea that retrieval module are useful to augment any deep AD method to enhance anomaly detection on tabular data.
Authors: Thomas Jiralerspong, Xiaoyin Chen, Yash More, Vedant Shah, Yoshua Bengio
Abstract: We propose a novel framework that leverages LLMs for full causal graph discovery. While previous LLM-based methods have used a pairwise query approach, this requires a quadratic number of queries which quickly becomes impractical for larger causal graphs. In contrast, the proposed framework uses a breadth-first search (BFS) approach which allows it to use only a linear number of queries. We also show that the proposed method can easily incorporate observational data when available, to improve performance. In addition to being more time and data-efficient, the proposed framework achieves state-of-the-art results on real-world causal graphs of varying sizes. The results demonstrate the effectiveness and efficiency of the proposed method in discovering causal relationships, showcasing its potential for broad applicability in causal graph discovery tasks across different domains.
Authors: Yuchen Zeng, Wonjun Kang, Yicong Chen, Hyung Il Koo, Kangwook Lee
Abstract: The evolution from Large Language Models (LLMs) to Multimodal Large Language Models (MLLMs) has spurred research into extending In-Context Learning (ICL) to its multimodal counterpart. Existing such studies have primarily concentrated on image-to-text ICL. However, the Text-to-Image ICL (T2I-ICL), with its unique characteristics and potential applications, remains underexplored. To address this gap, we formally define the task of T2I-ICL and present CoBSAT, the first T2I-ICL benchmark dataset, encompassing ten tasks. Utilizing our dataset to benchmark six state-of-the-art MLLMs, we uncover considerable difficulties MLLMs encounter in solving T2I-ICL. We identify the primary challenges as the inherent complexity of multimodality and image generation, and show that strategies such as fine-tuning and Chain-of-Thought prompting help to mitigate these difficulties, leading to notable improvements in performance. Our code and dataset are available at https://github.com/UW-Madison-Lee-Lab/CoBSAT.
Authors: Tianle Zhang, Yuchen Zhang, Kun Wang, Kai Wang, Beining Yang, Kaipeng Zhang, Wenqi Shao, Ping Liu, Joey Tianyi Zhou, Yang You
Abstract: Training on large-scale graphs has achieved remarkable results in graph representation learning, but its cost and storage have raised growing concerns. As one of the most promising directions, graph condensation methods address these issues by employing gradient matching, aiming to condense the full graph into a more concise yet information-rich synthetic set. Though encouraging, these strategies primarily emphasize matching directions of the gradients, which leads to deviations in the training trajectories. Such deviations are further magnified by the differences between the condensation and evaluation phases, culminating in accumulated errors, which detrimentally affect the performance of the condensed graphs. In light of this, we propose a novel graph condensation method named \textbf{C}raf\textbf{T}ing \textbf{R}ationa\textbf{L} trajectory (\textbf{CTRL}), which offers an optimized starting point closer to the original dataset's feature distribution and a more refined strategy for gradient matching. Theoretically, CTRL can effectively neutralize the impact of accumulated errors on the performance of condensed graphs. We provide extensive experiments on various graph datasets and downstream tasks to support the effectiveness of CTRL. Code is released at https://github.com/NUS-HPC-AI-Lab/CTRL.
Authors: Shipra Agrawal, Wei Tang
Abstract: We consider a dynamic pricing problem where customer response to the current price is impacted by the customer price expectation, aka reference price. We study a simple and novel reference price mechanism where reference price is the average of the past prices offered by the seller. As opposed to the more commonly studied exponential smoothing mechanism, in our reference price mechanism the prices offered by seller have a longer term effect on the future customer expectations. We show that under this mechanism, a markdown policy is near-optimal irrespective of the parameters of the model. This matches the common intuition that a seller may be better off by starting with a higher price and then decreasing it, as the customers feel like they are getting bargains on items that are ordinarily more expensive. For linear demand models, we also provide a detailed characterization of the near-optimal markdown policy along with an efficient way of computing it. We then consider a more challenging dynamic pricing and learning problem, where the demand model parameters are apriori unknown, and the seller needs to learn them online from the customers' responses to the offered prices while simultaneously optimizing revenue. The objective is to minimize regret, i.e., the $T$-round revenue loss compared to a clairvoyant optimal policy. This task essentially amounts to learning a non-stationary optimal policy in a time-variant Markov Decision Process (MDP). For linear demand models, we provide an efficient learning algorithm with an optimal $\tilde{O}(\sqrt{T})$ regret upper bound.
Authors: Heli Ben-Hamu, Omri Puny, Itai Gat, Brian Karrer, Uriel Singer, Yaron Lipman
Abstract: Taming the generation outcome of state of the art Diffusion and Flow-Matching (FM) models without having to re-train a task-specific model unlocks a powerful tool for solving inverse problems, conditional generation, and controlled generation in general. In this work we introduce D-Flow, a simple framework for controlling the generation process by differentiating through the flow, optimizing for the source (noise) point. We motivate this framework by our key observation stating that for Diffusion/FM models trained with Gaussian probability paths, differentiating through the generation process projects gradient on the data manifold, implicitly injecting the prior into the optimization process. We validate our framework on linear and non-linear controlled generation problems including: image and audio inverse problems and conditional molecule generation reaching state of the art performance across all.
Authors: Jules Berman, Benjamin Peherstorfer
Abstract: This work introduces reduced models based on Continuous Low Rank Adaptation (CoLoRA) that pre-train neural networks for a given partial differential equation and then continuously adapt low-rank weights in time to rapidly predict the evolution of solution fields at new physics parameters and new initial conditions. The adaptation can be either purely data-driven or via an equation-driven variational approach that provides Galerkin-optimal approximations. Because CoLoRA approximates solution fields locally in time, the rank of the weights can be kept small, which means that only few training trajectories are required offline so that CoLoRA is well suited for data-scarce regimes. Predictions with CoLoRA are orders of magnitude faster than with classical methods and their accuracy and parameter efficiency is higher compared to other neural network approaches.
Authors: Abhishek Dalvi, Vasant Honavar
Abstract: We introduce a novel method for transductive learning on graphs using hyperdimensional representations. The proposed approach encodes data samples using random projections into a very high-dimensional space (hyperdimensional or HD space for short). It obviates the need for expensive iterative training of the sort required by deep learning methods. Specifically, we propose a Hyperdimensional Graph Learning (HDGL) algorithm. HDGL leverages the \emph{injectivity} property of node representations of a family of Graph Neural Networks (GNNs) to map node features to the HD space and then uses HD operators such as bundling and binding to aggregate information from the local neighborhood of each node. The resulting latent node representations support both node classification and link prediction tasks, unlike typical deep learning methods, which often require separate models for these tasks. We report results of experiments using widely used benchmark datasets which demonstrate that, on the node classification task, HDGL is competitive with the SOTA GNN methods with respect to accuracy, at substantially reduced computational cost. Furthermore, HDGL is well-suited for class incremental learning where the model has to learn to effectively discriminate between a growing number of classes. Our experiments also show that the HD representation constructed by HDGL supports link prediction at accuracies comparable to that of DeepWalk and related methods, although it falls short of SOTA Graph Neural Network (GNN) methods that rely on computationally expensive iterative training. We conclude that HDGL offers a computationally efficient alternative to graph neural networks for node classification, especially in settings that call for class-incremental learning or in applications that demand high accuracy models at significantly lower computational cost and learning time than possible with the SOTA GNNs.
Authors: Eda Yilmaz, Hacer Yalim Keles
Abstract: We introduce Adversarial Sparse Teacher (AST), a robust defense method against distillation-based model stealing attacks. Our approach trains a teacher model using adversarial examples to produce sparse logit responses and increase the entropy of the output distribution. Typically, a model generates a peak in its output corresponding to its prediction. By leveraging adversarial examples, AST modifies the teacher model's original response, embedding a few altered logits into the output while keeping the primary response slightly higher. Concurrently, all remaining logits are elevated to further increase the output distribution's entropy. All these complex manipulations are performed using an optimization function with our proposed Exponential Predictive Divergence (EPD) loss function. EPD allows us to maintain higher entropy levels compared to traditional KL divergence, effectively confusing attackers. Experiments on CIFAR-10 and CIFAR-100 datasets demonstrate that AST outperforms state-of-the-art methods, providing effective defense against model stealing while preserving high accuracy. The source codes will be made publicly available here soon.
Authors: Alexandre Drouin, Maxime Gasse, Massimo Caccia, Issam H. Laradji, Manuel Del Verme, Tom Marty, L\'eo Boisvert, Megh Thakkar, Quentin Cappart, David Vazquez, Nicolas Chapados, Alexandre Lacoste
Abstract: We study the use of large language model-based agents for interacting with software via web browsers. Unlike prior work, we focus on measuring the agents' ability to perform tasks that span the typical daily work of knowledge workers utilizing enterprise software systems. To this end, we propose WorkArena, a remote-hosted benchmark of 33 tasks based on the widely-used ServiceNow platform. We also introduce BrowserGym, an environment for the design and evaluation of such agents, offering a rich set of actions as well as multimodal observations. Our empirical evaluation reveals that while current agents show promise on WorkArena, there remains a considerable gap towards achieving full task automation. Notably, our analysis uncovers a significant performance disparity between open and closed-source LLMs, highlighting a critical area for future exploration and development in the field.
Authors: Yi-Lun Liao, Tess Smidt, Muhammed Shuaibi, Abhishek Das
Abstract: Understanding the interactions of atoms such as forces in 3D atomistic systems is fundamental to many applications like molecular dynamics and catalyst design. However, simulating these interactions requires compute-intensive ab initio calculations and thus results in limited data for training neural networks. In this paper, we propose to use denoising non-equilibrium structures (DeNS) as an auxiliary task to better leverage training data and improve performance. For training with DeNS, we first corrupt a 3D structure by adding noise to its 3D coordinates and then predict the noise. Different from previous works on denoising, which are limited to equilibrium structures, the proposed method generalizes denoising to a much larger set of non-equilibrium structures. The main difference is that a non-equilibrium structure does not correspond to local energy minima and has non-zero forces, and therefore it can have many possible atomic positions compared to an equilibrium structure. This makes denoising non-equilibrium structures an ill-posed problem since the target of denoising is not uniquely defined. Our key insight is to additionally encode the forces of the original non-equilibrium structure to specify which non-equilibrium structure we are denoising. Concretely, given a corrupted non-equilibrium structure and the forces of the original one, we predict the non-equilibrium structure satisfying the input forces instead of any arbitrary structures. Since DeNS requires encoding forces, DeNS favors equivariant networks, which can easily incorporate forces and other higher-order tensors in node embeddings. We study the effectiveness of training equivariant networks with DeNS on OC20, OC22 and MD17 datasets and demonstrate that DeNS can achieve new state-of-the-art results on OC20 and OC22 and significantly improve training efficiency on MD17.
Authors: Haocheng Xi, Yuxiang Chen, Kang Zhao, Kai Jun Teh, Jianfei Chen, Jun Zhu
Abstract: Pretraining transformers are generally time-consuming. Fully quantized training (FQT) is a promising approach to speed up pretraining. However, most FQT methods adopt a quantize-compute-dequantize procedure, which often leads to suboptimal speedup and significant performance degradation when used in transformers due to the high memory access overheads and low-precision computations. In this work, we propose Jetfire, an efficient and accurate INT8 training method specific to transformers. Our method features an INT8 data flow to optimize memory access and a per-block quantization method to maintain the accuracy of pretrained transformers. Extensive experiments demonstrate that our INT8 FQT method achieves comparable accuracy to the FP16 training baseline and outperforms the existing INT8 training works for transformers. Moreover, for a standard transformer block, our method offers an end-to-end training speedup of 1.42x and a 1.49x memory reduction compared to the FP16 baseline. Our code is open sourced at https://github.com/thu-ml/Jetfire-INT8Training.
Authors: Daniel Zhengyu Huang, Jiaoyang Huang, Zhengjiang Lin
Abstract: Score-based generative models have emerged as a powerful approach for sampling high-dimensional probability distributions. Despite their effectiveness, their theoretical underpinnings remain relatively underdeveloped. In this work, we study the convergence properties of deterministic samplers based on probability flow ODEs from both theoretical and numerical perspectives. Assuming access to $L^2$-accurate estimates of the score function, we prove the total variation between the target and the generated data distributions can be bounded above by $\mathcal{O}(d^{3/4}\delta^{1/2})$ in the continuous time level, where $d$ denotes the data dimension and $\delta$ represents the $L^2$-score matching error. For practical implementations using a $p$-th order Runge-Kutta integrator with step size $h$, we establish error bounds of $\mathcal{O}(d^{3/4}\delta^{1/2} + d\cdot(dh)^p)$ at the discrete level. Finally, we present numerical studies on problems up to 128 dimensions to verify our theory.
Authors: Han Zhong, Guhao Feng, Wei Xiong, Xinle Cheng, Li Zhao, Di He, Jiang Bian, Liwei Wang
Abstract: In the classical Reinforcement Learning from Human Feedback (RLHF) framework, Proximal Policy Optimization (PPO) is employed to learn from sparse, sentence-level rewards -- a challenging scenario in traditional deep reinforcement learning. Despite the great successes of PPO in the alignment of state-of-the-art closed-source large language models (LLMs), its open-source implementation is still largely sub-optimal, as widely reported by numerous research studies. To address these issues, we introduce a framework that models RLHF problems as a Markov decision process (MDP), enabling the capture of fine-grained token-wise information. Furthermore, we provide theoretical insights that demonstrate the superiority of our MDP framework over the previous sentence-level bandit formulation. Under this framework, we introduce an algorithm, dubbed as Reinforced Token Optimization (\texttt{RTO}), which learns the token-wise reward function from preference data and performs policy optimization based on this learned token-wise reward signal. Theoretically, \texttt{RTO} is proven to have the capability of finding the near-optimal policy sample-efficiently. For its practical implementation, \texttt{RTO} innovatively integrates Direct Preference Optimization (DPO) and PPO. DPO, originally derived from sparse sentence rewards, surprisingly provides us with a token-wise characterization of response quality, which is seamlessly incorporated into our subsequent PPO training stage. Extensive real-world alignment experiments verify the effectiveness of the proposed approach.
Authors: Mustafa Cavus, Przemys{\l}aw Biecek
Abstract: Predictive models may generate biased predictions when classifying imbalanced datasets. This happens when the model favors the majority class, leading to low performance in accurately predicting the minority class. To address this issue, balancing or resampling methods are critical data-centric AI approaches in the modeling process to improve prediction performance. However, there have been debates and questions about the functionality of these methods in recent years. In particular, many candidate models may exhibit very similar predictive performance, called the Rashomon effect, in model selection, and they may even produce different predictions for the same observations. Selecting one of these models without considering the predictive multiplicity -- which is the case of yielding conflicting models' predictions for any sample -- can result in blind selection. In this paper, the impact of balancing methods on predictive multiplicity is examined using the Rashomon effect. It is crucial because the blind model selection in data-centric AI is risky from a set of approximately equally accurate models. This may lead to severe problems in model selection, validation, and explanation. To tackle this matter, we conducted real dataset experiments to observe the impact of balancing methods on predictive multiplicity through the Rashomon effect by using a newly proposed metric obscurity in addition to the existing ones: ambiguity and discrepancy. Our findings showed that balancing methods inflate the predictive multiplicity and yield varying results. To monitor the trade-off between the prediction performance and predictive multiplicity for conducting the modeling process responsibly, we proposed using the extended version of the performance-gain plot when balancing the training data.
Authors: Albert Bou, Morgan Thomas, Sebastian Dittert, Carles Navarro Ram\'irez, Maciej Majewski, Ye Wang, Shivam Patel, Gary Tresadern, Mazen Ahmad, Vincent Moens, Woody Sherman, Simone Sciabola, Gianni De Fabritiis
Abstract: In recent years, reinforcement learning (RL) has emerged as a valuable tool in drug design, offering the potential to propose and optimize molecules with desired properties. However, striking a balance between capabilities, flexibility, reliability, and efficiency remains challenging due to the complexity of advanced RL algorithms and the significant reliance on specialized code. In this work, we introduce ACEGEN, a comprehensive and streamlined toolkit tailored for generative drug design, built using TorchRL, a modern RL library that offers thoroughly tested reusable components. We validate ACEGEN by benchmarking against other published generative modeling algorithms and show comparable or improved performance. We also show examples of ACEGEN applied in multiple drug discovery case studies. ACEGEN is accessible at \url{https://github.com/acellera/acegen-open} and available for use under the MIT license.
Authors: Ruihao Gong, Yang Yong, Shiqiao Gu, Yushi Huang, Chentao Lv, Yunchen Zhang, Xianglong Liu, Dacheng Tao
Abstract: Recent advancements in large language models (LLMs) are propelling us toward artificial general intelligence with their remarkable emergent abilities and reasoning capabilities. However, the substantial computational and memory requirements limit the widespread adoption. Quantization, a key compression technique, can effectively mitigate these demands by compressing and accelerating LLMs, albeit with potential risks to accuracy. Numerous studies have aimed to minimize the accuracy loss associated with quantization. However, their quantization configurations vary from each other and cannot be fairly compared. In this paper, we present LLMC, a plug-and-play compression toolkit, to fairly and systematically explore the impact of quantization. LLMC integrates dozens of algorithms, models, and hardwares, offering high extensibility from integer to floating-point quantization, from LLM to vision-language (VLM) model, from fixed-bit to mixed precision, and from quantization to sparsification. Powered by this versatile toolkit, our benchmark covers three key aspects: calibration data, algorithms (three strategies), and data formats, providing novel insights and detailed analyses for further research and practical guidance for users. Our toolkit is available at \href{LLMC}{https://github.com/ModelTC/llmc}.
Authors: Jie Xu, Karthikeyan Saravanan, Rogier van Dalen, Haaris Mehmood, David Tuckey, Mete Ozay
Abstract: Federated learning (FL) allows clients 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 federated learning 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: Mengge Du, Yuntian Chen, Zhongzheng Wang, Longfeng Nie, Dongxiao Zhang
Abstract: Equation discovery is aimed at directly extracting physical laws from data and has emerged as a pivotal research domain. Previous methods based on symbolic mathematics have achieved substantial advancements, but often require the design of implementation of complex algorithms. In this paper, we introduce a new framework that utilizes natural language-based prompts to guide large language models (LLMs) in automatically mining governing equations from data. Specifically, we first utilize the generation capability of LLMs to generate diverse equations in string form, and then evaluate the generated equations based on observations. In the optimization phase, we propose two alternately iterated strategies to optimize generated equations collaboratively. The first strategy is to take LLMs as a black-box optimizer and achieve equation self-improvement based on historical samples and their performance. The second strategy is to instruct LLMs to perform evolutionary operators for global search. Experiments are extensively conducted on both partial differential equations and ordinary differential equations. Results demonstrate that our framework can discover effective equations to reveal the underlying physical laws under various nonlinear dynamic systems. Further comparisons are made with state-of-the-art models, demonstrating good stability and usability. Our framework substantially lowers the barriers to learning and applying equation discovery techniques, demonstrating the application potential of LLMs in the field of knowledge discovery.
Authors: Minyoung Huh, Brian Cheung, Tongzhou Wang, Phillip Isola
Abstract: We argue that representations in AI models, particularly deep networks, are converging. First, we survey many examples of convergence in the literature: over time and across multiple domains, the ways by which different neural networks represent data are becoming more aligned. Next, we demonstrate convergence across data modalities: as vision models and language models get larger, they measure distance between datapoints in a more and more alike way. We hypothesize that this convergence is driving toward a shared statistical model of reality, akin to Plato's concept of an ideal reality. We term such a representation the platonic representation and discuss several possible selective pressures toward it. Finally, we discuss the implications of these trends, their limitations, and counterexamples to our analysis.
Authors: Hao He, Chao Li, Wolfgang Ganglberger, Kaileigh Gallagher, Rumen Hristov, Michail Ouroutzoglou, Haoqi Sun, Jimeng Sun, Brandon Westover, Dina Katabi
Abstract: The ability to assess sleep at home, capture sleep stages, and detect the occurrence of apnea (without on-body sensors) simply by analyzing the radio waves bouncing off people's bodies while they sleep is quite powerful. Such a capability would allow for longitudinal data collection in patients' homes, informing our understanding of sleep and its interaction with various diseases and their therapeutic responses, both in clinical trials and routine care. In this article, we develop an advanced machine learning algorithm for passively monitoring sleep and nocturnal breathing from radio waves reflected off people while asleep. Validation results in comparison with the gold standard (i.e., polysomnography) (n=849) demonstrate that the model captures the sleep hypnogram (with an accuracy of 81% for 30-second epochs categorized into Wake, Light Sleep, Deep Sleep, or REM), detects sleep apnea (AUROC = 0.88), and measures the patient's Apnea-Hypopnea Index (ICC=0.95; 95% CI = [0.93, 0.97]). Notably, the model exhibits equitable performance across race, sex, and age. Moreover, the model uncovers informative interactions between sleep stages and a range of diseases including neurological, psychiatric, cardiovascular, and immunological disorders. These findings not only hold promise for clinical practice and interventional trials but also underscore the significance of sleep as a fundamental component in understanding and managing various diseases.
Authors: Xuyang Ge, Fukang Zhu, Wentao Shu, Junxuan Wang, Zhengfu He, Xipeng Qiu
Abstract: Circuit analysis of any certain model behavior is a central task in mechanistic interpretability. We introduce our circuit discovery pipeline with Sparse Autoencoders (SAEs) and a variant called Transcoders. With these two modules inserted into the model, the model's computation graph with respect to OV and MLP circuits becomes strictly linear. Our methods do not require linear approximation to compute the causal effect of each node. This fine-grained graph identifies both end-to-end and local circuits accounting for either logits or intermediate features. We can scalably apply this pipeline with a technique called Hierarchical Attribution. We analyze three kinds of circuits in GPT-2 Small: bracket, induction, and Indirect Object Identification circuits. Our results reveal new findings underlying existing discoveries.
Authors: Dirk Tasche
Abstract: The purpose of class distribution estimation (also known as quantification) is to determine the values of the prior class probabilities in a test dataset without class label observations. A variety of methods to achieve this have been proposed in the literature, most of them based on the assumption that the distributions of the training and test data are related through prior probability shift (also known as label shift). Among these methods, Friedman's method has recently been found to perform relatively well both for binary and multi-class quantification. We discuss the properties of Friedman's method and another approach mentioned by Friedman (called DeBias method in the literature) in the context of a general framework for designing linear equation systems for class distribution estimation.
Authors: Ying He, Mingyang Niu, Jingyu Hua, Yunlong Mao, Xu Huang, Chen Li, Sheng Zhong
Abstract: Split Neural Network, as one of the most common architectures used in vertical federated learning, is popular in industry due to its privacy-preserving characteristics. In this architecture, the party holding the labels seeks cooperation from other parties to improve model performance due to insufficient feature data. Each of these participants has a self-defined bottom model to learn hidden representations from its own feature data and uploads the embedding vectors to the top model held by the label holder for final predictions. This design allows participants to conduct joint training without directly exchanging data. However, existing research points out that malicious participants may still infer label information from the uploaded embeddings, leading to privacy leakage. In this paper, we first propose an embedding extension attack manipulating embeddings to undermine existing defense strategies, which rely on constraining the correlation between the embeddings uploaded by participants and the labels. Subsequently, we propose a new label obfuscation defense strategy, called `LabObf', which randomly maps each original integer-valued label to multiple real-valued soft labels with values intertwined, significantly increasing the difficulty for attackers to infer the labels. We conduct experiments on four different types of datasets, and the results show that LabObf significantly reduces the attacker's success rate compared to raw models while maintaining desirable model accuracy.
Authors: Jianliang He, Siyu Chen, Fengzhuo Zhang, Zhuoran Yang
Abstract: In this work, from a theoretical lens, we aim to understand why large language model (LLM) empowered agents are able to solve decision-making problems in the physical world. To this end, consider a hierarchical reinforcement learning (RL) model where the LLM Planner and the Actor perform high-level task planning and low-level execution, respectively. Under this model, the LLM Planner navigates a partially observable Markov decision process (POMDP) by iteratively generating language-based subgoals via prompting. Under proper assumptions on the pretraining data, we prove that the pretrained LLM Planner effectively performs Bayesian aggregated imitation learning (BAIL) through in-context learning. Additionally, we highlight the necessity for exploration beyond the subgoals derived from BAIL by proving that naively executing the subgoals returned by LLM leads to a linear regret. As a remedy, we introduce an $\epsilon$-greedy exploration strategy to BAIL, which is proven to incur sublinear regret when the pretraining error is small. Finally, we extend our theoretical framework to include scenarios where the LLM Planner serves as a world model for inferring the transition model of the environment and to multi-agent settings, enabling coordination among multiple Actors.
Authors: Kazuhiro Hiraki, Shinichi Ishihara, Junnosuke Shino
Abstract: This study first derives a general and analytical expression of AFA (Additive Feature Attribution) in terms of the kernel in LIME (Local Interpretable Model-agnostic Explanations). Then, we propose some new AFAs that have appropriate properties of kernels or that coincide with the LS prenucleolus in cooperative game theory. We also revisit existing AFAs such as SHAP (SHapley Additive exPlanations) and re-examine the properties of their kernels.
Authors: Hiroaki Chiba-Okabe, Weijie J. Su
Abstract: The rapid progress of generative AI technology has sparked significant copyright concerns, leading to numerous lawsuits filed against AI developers. While some studies explore methods to mitigate copyright risks by steering the outputs of generative models away from those resembling copyrighted data, little attention has been paid to the question of how much of a resemblance is undesirable; more original or unique data are afforded stronger protection, and the threshold level of resemblance for constituting infringement correspondingly lower. Here, leveraging this principle, we propose a genericization method that modifies the outputs of a generative model to make them more generic and less likely to infringe copyright. To achieve this, we introduce a metric for quantifying the level of originality of data in a manner that is consistent with the legal framework. This metric can be practically estimated by drawing samples from a generative model, which is then used for the genericization process. Experiments demonstrate that our genericization method successfully modifies the output of a text-to-image generative model so that it produces more generic, copyright-compliant images.
Authors: Liv Gorton
Abstract: Recent work on sparse autoencoders (SAEs) has shown promise in extracting interpretable features from neural networks and addressing challenges with polysemantic neurons caused by superposition. In this paper, we apply SAEs to the early vision layers of InceptionV1, a well-studied convolutional neural network, with a focus on curve detectors. Our results demonstrate that SAEs can uncover new interpretable features not apparent from examining individual neurons, including additional curve detectors that fill in previous gaps. We also find that SAEs can decompose some polysemantic neurons into more monosemantic constituent features. These findings suggest SAEs are a valuable tool for understanding InceptionV1, and convolutional neural networks more generally.
Authors: Jeongwhan Choi, Sumin Park, Hyowon Wi, Sung-Bae Cho, Noseong Park
Abstract: Recent research in the field of graph neural network (GNN) has identified a critical issue known as "over-squashing," resulting from the bottleneck phenomenon in graph structures, which impedes the propagation of long-range information. Prior works have proposed a variety of graph rewiring concepts that aim at optimizing the spatial or spectral properties of graphs to promote the signal propagation. However, such approaches inevitably deteriorate the original graph topology, which may lead to a distortion of information flow. To address this, we introduce an expanded width-aware (PANDA) message passing, a new message passing paradigm where nodes with high centrality, a potential source of over-squashing, are selectively expanded in width to encapsulate the growing influx of signals from distant nodes. Experimental results show that our method outperforms existing rewiring methods, suggesting that selectively expanding the hidden state of nodes can be a compelling alternative to graph rewiring for addressing the over-squashing.
Authors: William Yang, Ye Zhu, Zhiwei Deng, Olga Russakovsky
Abstract: Dataset distillation has emerged as a strategy to overcome the hurdles associated with large datasets by learning a compact set of synthetic data that retains essential information from the original dataset. While distilled data can be used to train high performing models, little is understood about how the information is stored. In this study, we posit and answer three questions about the behavior, representativeness, and point-wise information content of distilled data. We reveal distilled data cannot serve as a substitute for real data during training outside the standard evaluation setting for dataset distillation. Additionally, the distillation process retains high task performance by compressing information related to the early training dynamics of real models. Finally, we provide an framework for interpreting distilled data and reveal that individual distilled data points contain meaningful semantic information. This investigation sheds light on the intricate nature of distilled data, providing a better understanding on how they can be effectively utilized.
Authors: Vincent Cohen-Addad, Tommaso d'Orsi, Alessandro Epasto, Vahab Mirrokni, Peilin Zhong
Abstract: We revisit the input perturbations framework for differential privacy where noise is added to the input $A\in \mathcal{S}$ and the result is then projected back to the space of admissible datasets $\mathcal{S}$. Through this framework, we first design novel efficient algorithms to privately release pair-wise cosine similarities. Second, we derive a novel algorithm to compute $k$-way marginal queries over $n$ features. Prior work could achieve comparable guarantees only for $k$ even. Furthermore, we extend our results to $t$-sparse datasets, where our efficient algorithms yields novel, stronger guarantees whenever $t\le n^{5/6}/\log n\,.$ Finally, we provide a theoretical perspective on why \textit{fast} input perturbation algorithms works well in practice. The key technical ingredients behind our results are tight sum-of-squares certificates upper bounding the Gaussian complexity of sets of solutions.
Authors: Haitao Lin, Guojiang Zhao, Odin Zhang, Yufei Huang, Lirong Wu, Zicheng Liu, Siyuan Li, Cheng Tan, Zhifeng Gao, Stan Z. Li
Abstract: Structure-based drug design (SBDD) aims to generate potential drugs that can bind to a target protein and is greatly expedited by the aid of AI techniques in generative models. However, a lack of systematic understanding persists due to the diverse settings, complex implementation, difficult reproducibility, and task singularity. Firstly, the absence of standardization can lead to unfair comparisons and inconclusive insights. To address this dilemma, we propose CBGBench, a comprehensive benchmark for SBDD, that unifies the task as a generative heterogeneous graph completion, analogous to fill-in-the-blank of the 3D complex binding graph. By categorizing existing methods based on their attributes, CBGBench facilitates a modular and extensible framework that implements various cutting-edge methods. Secondly, a single task on \textit{de novo} molecule generation can hardly reflect their capabilities. To broaden the scope, we have adapted these models to a range of tasks essential in drug design, which are considered sub-tasks within the graph fill-in-the-blank tasks. These tasks include the generative designation of \textit{de novo} molecules, linkers, fragments, scaffolds, and sidechains, all conditioned on the structures of protein pockets. Our evaluations are conducted with fairness, encompassing comprehensive perspectives on interaction, chemical properties, geometry authenticity, and substructure validity. We further provide the pre-trained versions of the state-of-the-art models and deep insights with analysis from empirical studies. The codebase for CBGBench is publicly accessible at \url{https://github.com/Edapinenut/CBGBench}.
Authors: Zhicheng Liang, Yu Yang, Xiangyu Ke, Xiaokui Xiao, Yunjun Gao
Abstract: Recent years have witnessed a growing trend toward employing deep reinforcement learning (Deep-RL) to derive heuristics for combinatorial optimization (CO) problems on graphs. Maximum Coverage Problem (MCP) and its probabilistic variant on social networks, Influence Maximization (IM), have been particularly prominent in this line of research. In this paper, we present a comprehensive benchmark study that thoroughly investigates the effectiveness and efficiency of five recent Deep-RL methods for MCP and IM. These methods were published in top data science venues, namely S2V-DQN, Geometric-QN, GCOMB, RL4IM, and LeNSE. Our findings reveal that, across various scenarios, the Lazy Greedy algorithm consistently outperforms all Deep-RL methods for MCP. In the case of IM, theoretically sound algorithms like IMM and OPIM demonstrate superior performance compared to Deep-RL methods in most scenarios. Notably, we observe an abnormal phenomenon in IM problem where Deep-RL methods slightly outperform IMM and OPIM when the influence spread nearly does not increase as the budget increases. Furthermore, our experimental results highlight common issues when applying Deep-RL methods to MCP and IM in practical settings. Finally, we discuss potential avenues for improving Deep-RL methods. Our benchmark study sheds light on potential challenges in current deep reinforcement learning research for solving combinatorial optimization problems.
Authors: Shwai He, Guoheng Sun, Zheyu Shen, Ang Li
Abstract: Scaling Transformer-based large language models (LLMs) has demonstrated promising performance across various tasks. However, it also introduces redundant structures, posing challenges for real-world deployment. Despite some recognition of redundancy in LLMs, the variability of redundancy across different modules, such as MLP and Attention layers, is under-explored. In this work, we investigate the varying redundancy across different modules within Transformers, including Blocks, MLP, and Attention layers, using a similarity-based metric. This metric operates on the premise that redundant structures produce outputs highly similar to their inputs. Surprisingly, while attention layers are essential for transformers and distinguish them from other mainstream architectures, we found that a large proportion of attention layers exhibit excessively high similarity and can be safely pruned without degrading performance, leading to reduced memory and computation costs. Additionally, we further propose a method that jointly drops Attention and MLP layers, achieving improved performance and dropping ratios. Extensive experiments demonstrate the effectiveness of our methods, e.g., Llama-3-70B maintains comparable performance even after pruning half of the attention layers. Our findings provide valuable insights for future network architecture design. The code is released at: \url{https://github.com/Shwai-He/LLM-Drop}.
Authors: Peter Balazs, Martin Ehler, Daniel Haider
Abstract: Injectivity is the defining property of a mapping that ensures no information is lost and any input can be perfectly reconstructed from its output. By performing hard thresholding, the ReLU function naturally interferes with this property, making the injectivity analysis of ReLU-layers in neural networks a challenging yet intriguing task that has not yet been fully solved. This article establishes a frame theoretic perspective to approach this problem. The main objective is to develop the most general characterization of the injectivity behavior of ReLU-layers in terms of all three involved ingredients: (i) the weights, (ii) the bias, and (iii) the domain where the data is drawn from. Maintaining a focus on practical applications, we limit our attention to bounded domains and present two methods for numerically approximating a maximal bias for given weights and data domains. These methods provide sufficient conditions for the injectivity of a ReLU-layer on those domains and yield a novel practical methodology for studying the information loss in ReLU layers. Finally, we derive explicit reconstruction formulas based on the duality concept from frame theory.
Authors: Tobias Sebastian Finn, Charlotte Durand, Alban Farchi, Marc Bocquet, Julien Brajard
Abstract: We make the first steps towards diffusion models for unconditional generation of multivariate and Arctic-wide sea-ice states. While targeting to reduce the computational costs by diffusion in latent space, latent diffusion models also offer the possibility to integrate physical knowledge into the generation process. We tailor latent diffusion models to sea-ice physics with a censored Gaussian distribution in data space to generate data that follows the physical bounds of the modelled variables. Our latent diffusion models reach similar scores as the diffusion model trained in data space, but they smooth the generated fields as caused by the latent mapping. While enforcing physical bounds cannot reduce the smoothing, it improves the representation of the marginal ice zone. Therefore, for large-scale Earth system modelling, latent diffusion models can have many advantages compared to diffusion in data space if the significant barrier of smoothing can be resolved.
Authors: Jonas Ngnaw\'e, Sabyasachi Sahoo, Yann Pequignot, Fr\'ed\'eric Precioso, Christian Gagn\'e
Abstract: Despite extensive research on adversarial training strategies to improve robustness, the decisions of even the most robust deep learning models can still be quite sensitive to imperceptible perturbations, creating serious risks when deploying them for high-stakes real-world applications. While detecting such cases may be critical, evaluating a model's vulnerability at a per-instance level using adversarial attacks is computationally too intensive and unsuitable for real-time deployment scenarios. The input space margin is the exact score to detect non-robust samples and is intractable for deep neural networks. This paper introduces the concept of margin consistency -- a property that links the input space margins and the logit margins in robust models -- for efficient detection of vulnerable samples. First, we establish that margin consistency is a necessary and sufficient condition to use a model's logit margin as a score for identifying non-robust samples. Next, through comprehensive empirical analysis of various robustly trained models on CIFAR10 and CIFAR100 datasets, we show that they indicate strong margin consistency with a strong correlation between their input space margins and the logit margins. Then, we show that we can effectively use the logit margin to confidently detect brittle decisions with such models and accurately estimate robust accuracy on an arbitrarily large test set by estimating the input margins only on a small subset. Finally, we address cases where the model is not sufficiently margin-consistent by learning a pseudo-margin from the feature representation. Our findings highlight the potential of leveraging deep representations to efficiently assess adversarial vulnerability in deployment scenarios.
Authors: Isaac Ong, Amjad Almahairi, Vincent Wu, Wei-Lin Chiang, Tianhao Wu, Joseph E. Gonzalez, M Waleed Kadous, Ion Stoica
Abstract: Large language models (LLMs) exhibit impressive capabilities across a wide range of tasks, yet the choice of which model to use often involves a trade-off between performance and cost. More powerful models, though effective, come with higher expenses, while less capable models are more cost-effective. To address this dilemma, we propose several efficient router models that dynamically select between a stronger and a weaker LLM during inference, aiming to optimize the balance between cost and response quality. We develop a training framework for these routers leveraging human preference data and data augmentation techniques to enhance performance. Our evaluation on widely-recognized benchmarks shows that our approach significantly reduces costs-by over 2 times in certain cases-without compromising the quality of responses. Interestingly, our router models also demonstrate significant transfer learning capabilities, maintaining their performance even when the strong and weak models are changed at test time. This highlights the potential of these routers to provide a cost-effective yet high-performance solution for deploying LLMs.
Authors: Oliver R. A. Dunbar, Nicholas H. Nelsen, Maya Mutic
Abstract: Randomized algorithms exploit stochasticity to reduce computational complexity. One important example is random feature regression (RFR) that accelerates Gaussian process regression (GPR). RFR approximates an unknown function with a random neural network whose hidden weights and biases are sampled from a probability distribution. Only the final output layer is fit to data. In randomized algorithms like RFR, the hyperparameters that characterize the sampling distribution greatly impact performance, yet are not directly accessible from samples. This makes optimization of hyperparameters via standard (gradient-based) optimization tools inapplicable. Inspired by Bayesian ideas from GPR, this paper introduces a random objective function that is tailored for hyperparameter tuning of vector-valued random features. The objective is minimized with ensemble Kalman inversion (EKI). EKI is a gradient-free particle-based optimizer that is scalable to high-dimensions and robust to randomness in objective functions. A numerical study showcases the new black-box methodology to learn hyperparameter distributions in several problems that are sensitive to the hyperparameter selection: two global sensitivity analyses, integrating a chaotic dynamical system, and solving a Bayesian inverse problem from atmospheric dynamics. The success of the proposed EKI-based algorithm for RFR suggests its potential for automated optimization of hyperparameters arising in other randomized algorithms.
Authors: Bj\"orn van Zwol, Ro Jefferson, Egon L. van den Broek
Abstract: Recent years have witnessed a growing call for renewed emphasis on neuroscience-inspired approaches in artificial intelligence research, under the banner of NeuroAI. A prime example of this is predictive coding networks (PCNs), based on the neuroscientific framework of predictive coding. This framework views the brain as a hierarchical Bayesian inference model that minimizes prediction errors through feedback connections. Unlike traditional neural networks trained with backpropagation (BP), PCNs utilize inference learning (IL), a more biologically plausible algorithm that explains patterns of neural activity that BP cannot. Historically, IL has been more computationally intensive, but recent advancements have demonstrated that it can achieve higher efficiency than BP with sufficient parallelization. Furthermore, PCNs can be mathematically considered a superset of traditional feedforward neural networks (FNNs), significantly extending the range of trainable architectures. As inherently probabilistic (graphical) latent variable models, PCNs provide a versatile framework for both supervised learning and unsupervised (generative) modeling that goes beyond traditional artificial neural networks. This work provides a comprehensive review and detailed formal specification of PCNs, particularly situating them within the context of modern ML methods. Additionally, we introduce a Python library (PRECO) for practical implementation. This positions PC as a promising framework for future ML innovations.
Authors: Mouad El Bouchattaoui
Abstract: Meta-learning, or "learning to learn," is a subfield of machine learning where the goal is to develop models and algorithms that can learn from various tasks and improve their learning process over time. Unlike traditional machine learning methods focusing on learning a specific task, meta-learning aims to leverage experience from previous tasks to enhance future learning. This approach is particularly beneficial in scenarios where the available data for a new task is limited, but there exists abundant data from related tasks. By extracting and utilizing the underlying structure and patterns across these tasks, meta-learning algorithms can achieve faster convergence and better performance with fewer data. The following notes are mainly inspired from \cite{vanschoren2018meta}, \cite{baxter2019learning}, and \cite{maurer2005algorithmic}.
Authors: Nathan Gavenski, Juarez Monteiro, Felipe Meneguzzi, Michael Luck, Odinaldo Rodrigues
Abstract: Some imitation learning methods combine behavioural cloning with self-supervision to infer actions from state pairs. However, most rely on a large number of expert trajectories to increase generalisation and human intervention to capture key aspects of the problem, such as domain constraints. In this paper, we propose Continuous Imitation Learning from Observation (CILO), a new method augmenting imitation learning with two important features: (i) exploration, allowing for more diverse state transitions, requiring less expert trajectories and resulting in fewer training iterations; and (ii) path signatures, allowing for automatic encoding of constraints, through the creation of non-parametric representations of agents and expert trajectories. We compared CILO with a baseline and two leading imitation learning methods in five environments. It had the best overall performance of all methods in all environments, outperforming the expert in two of them.
Authors: Nick John Eliopoulos, Purvish Jajal, James Davis, Gaowen Liu, George K. Thiravathukal, Yung-Hsiang Lu
Abstract: This paper investigates how to efficiently deploy vision transformers on edge devices. Recent methods reduce the latency of transformer neural networks by removing or merging tokens, with small accuracy degradation. However, these methods are not designed with edge device deployment in mind: jthey do not leverage information about the latency vs. workload trends to improve efficiency. First, we show the latency-workload size relationship is nonlinear for certain workload sizes. We consider this relationship to create a token pruning schedule. Second, we demonstrate a training-free, token pruning method utilizing this schedule. We show that for single batch inference, other methods increase latency by 18.6-30.3% with respect to baseline, while we can reduce it by 9%. For similar latency (within 5.2%) across devices we achieve 78.6%-84.5% ImageNet1K accuracy, while the state-of-the-art, Token Merging, achieves 45.8%-85.4%.
Authors: Cameron Allen, Aaron Kirtland, Ruo Yu Tao, Sam Lobel, Daniel Scott, Nicholas Petrocelli, Omer Gottesman, Ronald Parr, Michael L. Littman, George Konidaris
Abstract: Reinforcement learning algorithms typically rely on the assumption that the environment dynamics and value function can be expressed in terms of a Markovian state representation. However, when state information is only partially observable, how can an agent learn such a state representation, and how can it detect when it has found one? We introduce a metric that can accomplish both objectives, without requiring access to--or knowledge of--an underlying, unobservable state space. Our metric, the $\lambda$-discrepancy, is the difference between two distinct temporal difference (TD) value estimates, each computed using TD($\lambda$) with a different value of $\lambda$. Since TD($\lambda$=0) makes an implicit Markov assumption and TD($\lambda$=1) does not, a discrepancy between these estimates is a potential indicator of a non-Markovian state representation. Indeed, we prove that the $\lambda$-discrepancy is exactly zero for all Markov decision processes and almost always non-zero for a broad class of partially observable environments. We also demonstrate empirically that, once detected, minimizing the $\lambda$-discrepancy can help with learning a memory function to mitigate the corresponding partial observability. We then train a reinforcement learning agent that simultaneously constructs two recurrent value networks with different $\lambda$ parameters and minimizes the difference between them as an auxiliary loss. The approach scales to challenging partially observable domains, where the resulting agent frequently performs significantly better (and never performs worse) than a baseline recurrent agent with only a single value network.
Authors: Arya Fayyazi, Mehdi Kamal, Massoud Pedram
Abstract: This paper presents ARCO, an adaptive Multi-Agent Reinforcement Learning (MARL)-based co-optimizing compilation framework designed to enhance the efficiency of mapping machine learning (ML) models - such as Deep Neural Networks (DNNs) - onto diverse hardware platforms. The framework incorporates three specialized actor-critic agents within MARL, each dedicated to a distinct aspect of compilation/optimization at an abstract level: one agent focuses on hardware, while two agents focus on software optimizations. This integration results in a collaborative hardware/software co-optimization strategy that improves the precision and speed of DNN deployments. Concentrating on high-confidence configurations simplifies the search space and delivers superior performance compared to current optimization methods. The ARCO framework surpasses existing leading frameworks, achieving a throughput increase of up to 37.95% while reducing the optimization time by up to 42.2% across various DNNs.
Authors: Yuxing Tian, Yiyan Qi, Aiwen Jiang, Qi Huang, Jian Guo
Abstract: Continuous-Time Dynamic Graph (CTDG) precisely models evolving real-world relationships, drawing heightened interest in dynamic graph learning across academia and industry. However, existing CTDG models encounter challenges stemming from noise and limited historical data. Graph Data Augmentation (GDA) emerges as a critical solution, yet current approaches primarily focus on static graphs and struggle to effectively address the dynamics inherent in CTDGs. Moreover, these methods often demand substantial domain expertise for parameter tuning and lack theoretical guarantees for augmentation efficacy. To address these issues, we propose Conda, a novel latent diffusion-based GDA method tailored for CTDGs. Conda features a sandwich-like architecture, incorporating a Variational Auto-Encoder (VAE) and a conditional diffusion model, aimed at generating enhanced historical neighbor embeddings for target nodes. Unlike conventional diffusion models trained on entire graphs via pre-training, Conda requires historical neighbor sequence embeddings of target nodes for training, thus facilitating more targeted augmentation. We integrate Conda into the CTDG model and adopt an alternating training strategy to optimize performance. Extensive experimentation across six widely used real-world datasets showcases the consistent performance improvement of our approach, particularly in scenarios with limited historical data.
Authors: Musleh Alharthi, Ausif Mahmood
Abstract: In recent years, transformer-based models have gained prominence in multivariate long-term time series forecasting (LTSF), demonstrating significant advancements despite facing challenges such as high computational demands, difficulty in capturing temporal dynamics, and managing long-term dependencies. The emergence of LTSF-Linear, with its straightforward linear architecture, has notably outperformed transformer-based counterparts, prompting a reevaluation of the transformer's utility in time series forecasting. In response, this paper presents an adaptation of a recent architecture termed extended LSTM (xLSTM) for LTSF. xLSTM incorporates exponential gating and a revised memory structure with higher capacity that has good potential for LTSF. Our adopted architecture for LTSF termed as xLSTMTime surpasses current approaches. We compare xLSTMTime's performance against various state-of-the-art models across multiple real-world da-tasets, demonstrating superior forecasting capabilities. Our findings suggest that refined recurrent architectures can offer competitive alternatives to transformer-based models in LTSF tasks, po-tentially redefining the landscape of time series forecasting.
Authors: Yi Shen, Shao Gu
Abstract: Recently, interesting empirical phenomena known as Neural Collapse have been observed during the final phase of training deep neural networks for classification tasks. We examine this issue when the feature dimension d is equal to the number of classes K. We demonstrate that two popular unconstrained feature models are strict saddle functions, with every critical point being either a global minimum or a strict saddle point that can be exited using negative curvatures. The primary findings conclusively confirm the conjecture on the unconstrained feature models in previous articles.
Authors: GaoXiang Zhao, XiaoQiang Wang
Abstract: The field of long-term time series forecasting demands handling extensive look-back windows and long-range prediction steps, posing significant challenges for RNN-based methodologies. Among these, SegRNN, a robust RNN-driven model, has gained considerable attention in LTSF analysis for achieving state-of-the-art results while maintaining a remarkably streamlined architecture. Concurrently, the Mamba structure has demonstrated its advantages in small to medium-sized models due to its capability for information selection. This study introduces a variant of SegRNN that preprocesses information using a fine-tuned single-layer Mamba structure. Additionally, it incorporates implicit segmentation and residual structures into the model's encoding section to further reduce the inherent data iterative cycles of RNN architectures and implicitly integrate inter-channel correlations. This variant, named MSegRNN, utilizes the Mamba structure to select useful information, resulting in a transformed sequence. The linear-strategy-adapted derivative retains the superior memory efficiency of the original SegRNN while demonstrating enhanced performance. Empirical evaluations on real-world LTSF datasets demonstrate the superior performance of our model, thereby contributing to the advancement of LTSF methodologies.
Authors: Kim Hammar, Neil Dhir, Rolf Stadler
Abstract: The CAGE-2 challenge is considered a standard benchmark to compare methods for autonomous cyber defense. Current state-of-the-art methods evaluated against this benchmark are based on model-free (offline) reinforcement learning, which does not provide provably optimal defender strategies. We address this limitation and present a formal (causal) model of CAGE-2 together with a method that produces a provably optimal defender strategy, which we call Causal Partially Observable Monte-Carlo Planning (C-POMCP). It has two key properties. First, it incorporates the causal structure of the target system, i.e., the causal relationships among the system variables. This structure allows for a significant reduction of the search space of defender strategies. Second, it is an online method that updates the defender strategy at each time step via tree search. Evaluations against the CAGE-2 benchmark show that C-POMCP achieves state-of-the-art performance with respect to effectiveness and is two orders of magnitude more efficient in computing time than the closest competitor method.
Authors: Seyma Gunonu, Gizem Altun, Mustafa Cavus
Abstract: The accuracy and understandability of bank failure prediction models are crucial. While interpretable models like logistic regression are favored for their explainability, complex models such as random forest, support vector machines, and deep learning offer higher predictive performance but lower explainability. These models, known as black boxes, make it difficult to derive actionable insights. To address this challenge, using counterfactual explanations is suggested. These explanations demonstrate how changes in input variables can alter the model output and suggest ways to mitigate bank failure risk. The key challenge lies in selecting the most effective method for generating useful counterfactuals, which should demonstrate validity, proximity, sparsity, and plausibility. The paper evaluates several counterfactual generation methods: WhatIf, Multi Objective, and Nearest Instance Counterfactual Explanation, and also explores resampling methods like undersampling, oversampling, SMOTE, and the cost sensitive approach to address data imbalance in bank failure prediction in the US. The results indicate that the Nearest Instance Counterfactual Explanation method yields higher quality counterfactual explanations, mainly using the cost sensitive approach. Overall, the Multi Objective Counterfactual and Nearest Instance Counterfactual Explanation methods outperform others regarding validity, proximity, and sparsity metrics, with the cost sensitive approach providing the most desirable counterfactual explanations. These findings highlight the variability in the performance of counterfactual generation methods across different balancing strategies and machine learning models, offering valuable strategies to enhance the utility of black box bank failure prediction models.
Authors: Jakub Adamczyk, Wojciech Czech
Abstract: We revisit the effectiveness of topological descriptors for molecular graph classification and design a simple, yet strong baseline. We demonstrate that a simple approach to feature engineering - employing histogram aggregation of edge descriptors and one-hot encoding for atomic numbers and bond types - when combined with a Random Forest classifier, can establish a strong baseline for Graph Neural Networks (GNNs). The novel algorithm, Molecular Topological Profile (MOLTOP), integrates Edge Betweenness Centrality, Adjusted Rand Index and SCAN Structural Similarity score. This approach proves to be remarkably competitive when compared to modern GNNs, while also being simple, fast, low-variance and hyperparameter-free. Our approach is rigorously tested on MoleculeNet datasets using fair evaluation protocol provided by Open Graph Benchmark. We additionally show out-of-domain generation capabilities on peptide classification task from Long Range Graph Benchmark. The evaluations across eleven benchmark datasets reveal MOLTOP's strong discriminative capabilities, surpassing the $1$-WL test and even $3$-WL test for some classes of graphs. Our conclusion is that descriptor-based baselines, such as the one we propose, are still crucial for accurately assessing advancements in the GNN domain.
Authors: Daniel Tan, David Chanin, Aengus Lynch, Dimitrios Kanoulas, Brooks Paige, Adria Garriga-Alonso, Robert Kirk
Abstract: Steering vectors (SVs) are a new approach to efficiently adjust language model behaviour at inference time by intervening on intermediate model activations. They have shown promise in terms of improving both capabilities and model alignment. However, the reliability and generalisation properties of this approach are unknown. In this work, we rigorously investigate these properties, and show that steering vectors have substantial limitations both in- and out-of-distribution. In-distribution, steerability is highly variable across different inputs. Depending on the concept, spurious biases can substantially contribute to how effective steering is for each input, presenting a challenge for the widespread use of steering vectors. Out-of-distribution, while steering vectors often generalise well, for several concepts they are brittle to reasonable changes in the prompt, resulting in them failing to generalise well. Overall, our findings show that while steering can work well in the right circumstances, there remain many technical difficulties of applying steering vectors to guide models' behaviour at scale.
Authors: Mohammed Ghaith Altarabichi
Abstract: We propose DropKAN (Dropout Kolmogorov-Arnold Networks) a regularization method that prevents co-adaptation of activation function weights in Kolmogorov-Arnold Networks (KANs). DropKAN operates by randomly masking some of the post-activations within the KANs computation graph, while scaling-up the retained post-activations. We show that this simple procedure that require minimal coding effort has a regularizing effect and consistently lead to better generalization of KANs. We analyze the adaptation of the standard Dropout with KANs and demonstrate that Dropout applied to KANs' neurons can lead to unpredictable behaviour in the feedforward pass. We carry an empirical study with real world Machine Learning datasets to validate our findings. Our results suggest that DropKAN is consistently a better alternative to using standard Dropout with KANs, and improves the generalization performance of KANs. Our implementation of DropKAN is available at: \url{https://github.com/Ghaith81/dropkan}.
Authors: Yingru Li, Jiawei Xu, Zhi-Quan Luo
Abstract: Foundation models often struggle with uncertainty when faced with new situations in online decision-making, necessitating scalable and efficient exploration to resolve this uncertainty. We introduce GPT-HyperAgent, an augmentation of GPT with HyperAgent for uncertainty-aware, scalable exploration in contextual bandits, a fundamental online decision problem involving natural language input. We prove that HyperAgent achieves fast incremental uncertainty estimation with $\tilde{O}(\log T)$ per-step computational complexity over $T$ periods under the linear realizable assumption. Our analysis demonstrates that HyperAgent's regret order matches that of exact Thompson sampling in linear contextual bandits, closing a significant theoretical gap in scalable exploration. Empirical results in real-world contextual bandit tasks, such as automated content moderation with human feedback, validate the practical effectiveness of GPT-HyperAgent for safety-critical decisions. Our code is open-sourced at \url{https://github.com/szrlee/GPT-HyperAgent/}.
Authors: Danielle Ensign, Adri\`a Garriga-Alonso
Abstract: How well will current interpretability techniques generalize to future models? A relevant case study is Mamba, a recent recurrent architecture with scaling comparable to Transformers. We adapt pre-Mamba techniques to Mamba and partially reverse-engineer the circuit responsible for the Indirect Object Identification (IOI) task. Our techniques provide evidence that 1) Layer 39 is a key bottleneck, 2) Convolutions in layer 39 shift names one position forward, and 3) The name entities are stored linearly in Layer 39's SSM. Finally, we adapt an automatic circuit discovery tool, positional Edge Attribution Patching, to identify a Mamba IOI circuit. Our contributions provide initial evidence that circuit-based mechanistic interpretability tools work well for the Mamba architecture.
Authors: Sanzo Miyazawa
Abstract: The inverse Potts problem to infer a Boltzmann distribution for homologous protein sequences from their single-site and pairwise amino acid frequencies recently attracts a great deal of attention in the studies of protein structure and evolution. We study regularization and learning methods and how to tune regularization parameters to correctly infer interactions in Boltzmann machine learning. Using $L_2$ regularization for fields, group $L_1$ for couplings is shown to be very effective for sparse couplings in comparison with $L_2$ and $L_1$. Two regularization parameters are tuned to yield equal values for both the sample and ensemble averages of evolutionary energy. Both averages smoothly change and converge, but their learning profiles are very different between learning methods. The Adam method is modified to make stepsize proportional to the gradient for sparse couplings. It is shown by first inferring interactions from protein sequences and then from Monte Carlo samples that the fields and couplings can be well recovered, but that recovering the pairwise correlations in the resolution of a total energy is harder for the natural proteins than for the protein-like sequences. Selective temperature for folding/structural constrains in protein evolution is also estimated.
Authors: Dan Li, Dariush Fooladivanda, Sonia Martinez
Abstract: This paper proposes a novel approach to construct data-driven online solutions to optimization problems (P) subject to a class of distributionally uncertain dynamical systems. The introduced framework allows for the simultaneous learning of distributional system uncertainty via a parameterized, control-dependent ambiguity set using a finite historical data set, and its use to make online decisions with probabilistic regret function bounds. Leveraging the merits of Machine Learning, the main technical approach relies on the theory of Distributional Robust Optimization (DRO), to hedge against uncertainty and provide less conservative results than standard Robust Optimization approaches. Starting from recent results that describe ambiguity sets via parameterized, and control-dependent empirical distributions as well as ambiguity radii, we first present a tractable reformulation of the corresponding optimization problem while maintaining the probabilistic guarantees. We then specialize these problems to the cases of 1) optimal one-stage control of distributionally uncertain nonlinear systems, and 2) resource allocation under distributional uncertainty. A novelty of this work is that it extends DRO to online optimization problems subject to a distributionally uncertain dynamical system constraint, handled via a control-dependent ambiguity set that leads to online-tractable optimization with probabilistic guarantees on regret bounds. Further, we introduce an online version of Nesterov's accelerated-gradient algorithm, and analyze its performance to solve this class of problems via dissipativity theory.
Authors: Minlong Peng, Zidi Xiong, Quang H. Nguyen, Mingming Sun, Khoa D. Doan, Ping Li
Abstract: By injecting a small number of poisoned samples into the training set, backdoor attacks aim to make the victim model produce designed outputs on any input injected with pre-designed backdoors. In order to achieve a high attack success rate using as few poisoned training samples as possible, most existing attack methods change the labels of the poisoned samples to the target class. This practice often results in severe over-fitting of the victim model over the backdoors, making the attack quite effective in output control but easier to be identified by human inspection or automatic defense algorithms. In this work, we proposed a label-smoothing strategy to overcome the over-fitting problem of these attack methods, obtaining a \textit{Label-Smoothed Backdoor Attack} (LSBA). In the LSBA, the label of the poisoned sample $\bm{x}$ will be changed to the target class with a probability of $p_n(\bm{x})$ instead of 100\%, and the value of $p_n(\bm{x})$ is specifically designed to make the prediction probability the target class be only slightly greater than those of the other classes. Empirical studies on several existing backdoor attacks show that our strategy can considerably improve the stealthiness of these attacks and, at the same time, achieve a high attack success rate. In addition, our strategy makes it able to manually control the prediction probability of the design output through manipulating the applied and activated number of LSBAs\footnote{Source code will be published at \url{https://github.com/v-mipeng/LabelSmoothedAttack.git}}.
Authors: Fenglin Liu, Bang Yang, Chenyu You, Xian Wu, Shen Ge, Zhangdaihong Liu, Xu Sun, Yang Yang, David A. Clifton
Abstract: Language models (LMs), including large language models (such as ChatGPT), have the potential to assist clinicians in generating various clinical notes. However, LMs are prone to produce ``hallucinations'', i.e., generated content that is not aligned with facts and knowledge. In this paper, we propose the Re$^3$Writer method with retrieval-augmented generation and knowledge-grounded reasoning to enable LMs to generate faithful clinical texts. We demonstrate the effectiveness of our method in generating patient discharge instructions. It requires the LMs not to only understand the patients' long clinical documents, i.e., the health records during hospitalization, but also to generate critical instructional information provided both to carers and to the patient at the time of discharge. The proposed Re$^3$Writer imitates the working patterns of physicians to first \textbf{re}trieve related working experience from historical instructions written by physicians, then \textbf{re}ason related medical knowledge. Finally, it \textbf{re}fines the retrieved working experience and reasoned medical knowledge to extract useful information, which is used to generate the discharge instructions for previously-unseen patients. Our experiments show that, using our method, the performance of five representative LMs can be substantially boosted across all metrics. Meanwhile, we show results from human evaluations to measure the effectiveness in terms of fluency, faithfulness, and comprehensiveness.
Authors: Valerio Guarrasi, Lorenzo Tronchin, Domenico Albano, Eliodoro Faiella, Deborah Fazzini, Domiziana Santucci, Paolo Soda
Abstract: We are witnessing a widespread adoption of artificial intelligence in healthcare. However, most of the advancements in deep learning in this area consider only unimodal data, neglecting other modalities. Their multimodal interpretation necessary for supporting diagnosis, prognosis and treatment decisions. In this work we present a deep architecture, which jointly learns modality reconstructions and sample classifications using tabular and imaging data. The explanation of the decision taken is computed by applying a latent shift that, simulates a counterfactual prediction revealing the features of each modality that contribute the most to the decision and a quantitative score indicating the modality importance. We validate our approach in the context of COVID-19 pandemic using the AIforCOVID dataset, which contains multimodal data for the early identification of patients at risk of severe outcome. The results show that the proposed method provides meaningful explanations without degrading the classification performance.
Authors: Bruce W. Lee, Jason Hyung-Jong Lee
Abstract: Traditional English readability formulas, or equations, were largely developed in the 20th century. Nonetheless, many researchers still rely on them for various NLP applications. This phenomenon is presumably due to the convenience and straightforwardness of readability formulas. In this work, we contribute to the NLP community by 1. introducing New English Readability Formula (NERF), 2. recalibrating the coefficients of old readability formulas (Flesch-Kincaid Grade Level, Fog Index, SMOG Index, Coleman-Liau Index, and Automated Readability Index), 3. evaluating the readability formulas, for use in text simplification studies and medical texts, and 4. developing a Python-based program for the wide application to various NLP projects.
Authors: Cencheng Shen, Youngser Park, Carey E. Priebe
Abstract: In this paper, we introduce a novel and computationally efficient method for vertex embedding, community detection, and community size determination. Our approach leverages a normalized one-hot graph encoder and a rank-based cluster size measure. Through extensive simulations, we demonstrate the excellent numerical performance of our proposed graph encoder ensemble algorithm.
Authors: Ning Liu, Benjamin Grimmer
Abstract: We consider feasibility and constrained optimization problems defined over smooth and/or strongly convex sets. These notions mirror their popular function counterparts but are much less explored in the first-order optimization literature. We propose new scalable, projection-free, accelerated first-order methods in these settings. Our methods avoid linear optimization or projection oracles, only using cheap one-dimensional linesearches and normal vector computations. Despite this, we derive optimal accelerated convergence guarantees of $O(1/T)$ for strongly convex problems, $O(1/T^2)$ for smooth problems, and accelerated linear convergence given both. Our algorithms and analysis are based on novel characterizations of the Minkowski gauge of smooth and/or strongly convex sets, which may be of independent interest: although the gauge is neither smooth nor strongly convex, we show the gauge squared inherits any structure present in the set.
Authors: Sukrut Rao, Moritz B\"ohle, Bernt Schiele
Abstract: Deep neural networks are very successful on many vision tasks, but hard to interpret due to their black box nature. To overcome this, various post-hoc attribution methods have been proposed to identify image regions most influential to the models' decisions. Evaluating such methods is challenging since no ground truth attributions exist. We thus propose three novel evaluation schemes to more reliably measure the faithfulness of those methods, to make comparisons between them more fair, and to make visual inspection more systematic. To address faithfulness, we propose a novel evaluation setting (DiFull) in which we carefully control which parts of the input can influence the output in order to distinguish possible from impossible attributions. To address fairness, we note that different methods are applied at different layers, which skews any comparison, and so evaluate all methods on the same layers (ML-Att) and discuss how this impacts their performance on quantitative metrics. For more systematic visualizations, we propose a scheme (AggAtt) to qualitatively evaluate the methods on complete datasets. We use these evaluation schemes to study strengths and shortcomings of some widely used attribution methods over a wide range of models. Finally, we propose a post-processing smoothing step that significantly improves the performance of some attribution methods, and discuss its applicability.
Authors: Sukrut Rao, Moritz B\"ohle, Amin Parchami-Araghi, Bernt Schiele
Abstract: Despite being highly performant, deep neural networks might base their decisions on features that spuriously correlate with the provided labels, thus hurting generalization. To mitigate this, 'model guidance' has recently gained popularity, i.e. the idea of regularizing the models' explanations to ensure that they are "right for the right reasons". While various techniques to achieve such model guidance have been proposed, experimental validation of these approaches has thus far been limited to relatively simple and / or synthetic datasets. To better understand the effectiveness of the various design choices that have been explored in the context of model guidance, in this work we conduct an in-depth evaluation across various loss functions, attribution methods, models, and 'guidance depths' on the PASCAL VOC 2007 and MS COCO 2014 datasets. As annotation costs for model guidance can limit its applicability, we also place a particular focus on efficiency. Specifically, we guide the models via bounding box annotations, which are much cheaper to obtain than the commonly used segmentation masks, and evaluate the robustness of model guidance under limited (e.g. with only 1% of annotated images) or overly coarse annotations. Further, we propose using the EPG score as an additional evaluation metric and loss function ('Energy loss'). We show that optimizing for the Energy loss leads to models that exhibit a distinct focus on object-specific features, despite only using bounding box annotations that also include background regions. Lastly, we show that such model guidance can improve generalization under distribution shifts. Code available at: https://github.com/sukrutrao/Model-Guidance.
Authors: Jaemin Cho, Linjie Li, Zhengyuan Yang, Zhe Gan, Lijuan Wang, Mohit Bansal
Abstract: Spatial control is a core capability in controllable image generation. Advancements in layout-guided image generation have shown promising results on in-distribution (ID) datasets with similar spatial configurations. However, it is unclear how these models perform when facing out-of-distribution (OOD) samples with arbitrary, unseen layouts. In this paper, we propose LayoutBench, a diagnostic benchmark for layout-guided image generation that examines four categories of spatial control skills: number, position, size, and shape. We benchmark two recent representative layout-guided image generation methods and observe that the good ID layout control may not generalize well to arbitrary layouts in the wild (e.g., objects at the boundary). Next, we propose IterInpaint, a new baseline that generates foreground and background regions step-by-step via inpainting, demonstrating stronger generalizability than existing models on OOD layouts in LayoutBench. We perform quantitative and qualitative evaluation and fine-grained analysis on the four LayoutBench skills to pinpoint the weaknesses of existing models. We show comprehensive ablation studies on IterInpaint, including training task ratio, crop&paste vs. repaint, and generation order. Lastly, we evaluate the zero-shot performance of different pretrained layout-guided image generation models on LayoutBench-COCO, our new benchmark for OOD layouts with real objects, where our IterInpaint consistently outperforms SOTA baselines in all four splits. Project website: https://layoutbench.github.io
Authors: Jun Zhang, Xiaohan Lin, Weinan E, Yi Qin Gao
Abstract: Multiscale molecular modeling is widely applied in scientific research of molecular properties over large time and length scales. Two specific challenges are commonly present in multiscale modeling, provided that information between the coarse and fine representations of molecules needs to be properly exchanged: One is to construct coarse grained models by passing information from the fine to coarse levels; the other is to restore finer molecular details given coarse grained configurations. Although these two problems are commonly addressed independently, in this work, we present a theory connecting them, and develop a methodology called Cycle Coarse Graining (CCG) to solve both problems in a unified manner. In CCG, reconstruction can be achieved via a tractable deep generative model, allowing retrieval of fine details from coarse-grained simulations. The reconstruction in turn delivers better coarse-grained models which are informed of the fine-grained physics, and enables calculation of the free energies in a rare-event-free manner. CCG thus provides a systematic way for multiscale molecular modeling, where the finer details of coarse-grained simulations can be efficiently retrieved, and the coarse-grained models can be improved consistently.
Authors: Zhan Yu, Jun Fan, Zhongjie Shi, Ding-Xuan Zhou
Abstract: In recent years, different types of distributed and parallel learning schemes have received increasing attention for their strong advantages in handling large-scale data information. In the information era, to face the big data challenges {that} stem from functional data analysis very recently, we propose a novel distributed gradient descent functional learning (DGDFL) algorithm to tackle functional data across numerous local machines (processors) in the framework of reproducing kernel Hilbert space. Based on integral operator approaches, we provide the first theoretical understanding of the DGDFL algorithm in many different aspects of the literature. On the way of understanding DGDFL, firstly, a data-based gradient descent functional learning (GDFL) algorithm associated with a single-machine model is proposed and comprehensively studied. Under mild conditions, confidence-based optimal learning rates of DGDFL are obtained without the saturation boundary on the regularity index suffered in previous works in functional regression. We further provide a semi-supervised DGDFL approach to weaken the restriction on the maximal number of local machines to ensure optimal rates. To our best knowledge, the DGDFL provides the first divide-and-conquer iterative training approach to functional learning based on data samples of intrinsically infinite-dimensional random functions (functional covariates) and enriches the methodologies for functional data analysis.
Authors: Alexander Shevtsov, Despoina Antonakaki, Ioannis Lamprou, Polyvios Pratikakis, Sotiris Ioannidis
Abstract: Twitter, as one of the most popular social networks, provides a platform for communication and online discourse. Unfortunately, it has also become a target for bots and fake accounts, resulting in the spread of false information and manipulation. This paper introduces a semi-automatic machine learning pipeline (SAMLP) designed to address the challenges correlated with machine learning model development. Through this pipeline, we develop a comprehensive bot detection model named BotArtist, based on user profile features. SAMLP leverages nine distinct publicly available datasets to train the BotArtist model. To assess BotArtist's performance against current state-of-the-art solutions, we select 35 existing Twitter bot detection methods, each utilizing a diverse range of features. Our comparative evaluation of BotArtist and these existing methods, conducted across nine public datasets under standardized conditions, reveals that the proposed model outperforms existing solutions by almost 10%, in terms of F1-score, achieving an average score of 83.19 and 68.5 over specific and general approaches respectively. As a result of this research, we provide a dataset of the extracted features combined with BotArtist predictions over the 10.929.533 Twitter user profiles, collected via Twitter API during the 2022 Russo-Ukrainian War, over a 16-month period. This dataset was created in collaboration with [Shevtsov et al., 2022a] where the original authors share anonymized tweets on the discussion of the Russo-Ukrainian war with a total amount of 127.275.386 tweets. The combination of the existing text dataset and the provided labeled bot and human profiles will allow for the future development of a more advanced bot detection large language model in the post-Twitter API era.
Authors: Pierre Clavier, Tom Huix, Alain Durmus
Abstract: In this paper, we introduce and analyze a variant of the Thompson sampling (TS) algorithm for contextual bandits. At each round, traditional TS requires samples from the current posterior distribution, which is usually intractable. To circumvent this issue, approximate inference techniques can be used and provide samples with distribution close to the posteriors. However, current approximate techniques yield to either poor estimation (Laplace approximation) or can be computationally expensive (MCMC methods, Ensemble sampling...). In this paper, we propose a new algorithm, Varational Inference Thompson sampling VITS, based on Gaussian Variational Inference. This scheme provides powerful posterior approximations which are easy to sample from, and is computationally efficient, making it an ideal choice for TS. In addition, we show that VITS achieves a sub-linear regret bound of the same order in the dimension and number of round as traditional TS for linear contextual bandit. Finally, we demonstrate experimentally the effectiveness of VITS on both synthetic and real world datasets.
Authors: Thomas T. C. K. Zhang, Leonardo F. Toso, James Anderson, Nikolai Matni
Abstract: A powerful concept behind much of the recent progress in machine learning is the extraction of common features across data from heterogeneous sources or tasks. Intuitively, using all of one's data to learn a common representation function benefits both computational effort and statistical generalization by leaving a smaller number of parameters to fine-tune on a given task. Toward theoretically grounding these merits, we propose a general setting of recovering linear operators $M$ from noisy vector measurements $y = Mx + w$, where the covariates $x$ may be both non-i.i.d. and non-isotropic. We demonstrate that existing isotropy-agnostic representation learning approaches incur biases on the representation update, which causes the scaling of the noise terms to lose favorable dependence on the number of source tasks. This in turn can cause the sample complexity of representation learning to be bottlenecked by the single-task data size. We introduce an adaptation, $\texttt{De-bias & Feature-Whiten}$ ($\texttt{DFW}$), of the popular alternating minimization-descent scheme proposed independently in Collins et al., (2021) and Nayer and Vaswani (2022), and establish linear convergence to the optimal representation with noise level scaling down with the $\textit{total}$ source data size. This leads to generalization bounds on the same order as an oracle empirical risk minimizer. We verify the vital importance of $\texttt{DFW}$ on various numerical simulations. In particular, we show that vanilla alternating-minimization descent fails catastrophically even for iid, but mildly non-isotropic data. Our analysis unifies and generalizes prior work, and provides a flexible framework for a wider range of applications, such as in controls and dynamical systems.
Authors: Parikshit Pareek, Deepjyoti Deka, Sidhant Misra
Abstract: This paper presents a graph-structured Gaussian process (GP) model for data-driven risk assessment of critical voltage constraints. The proposed GP is based on a novel kernel, named the vertex-degree kernel (VDK), that decomposes the voltage-load relationship based on the network graph. To estimate the GP efficiently, we propose a novel active learning scheme that leverages the additive structure of VDK. Further, we prove a probabilistic bound on the error in risk estimation using VDK-GP model that demonstrates that it is statistically comparable to using standard AC power flow (AC-PF), but does not require computing a large number of ACPF solutions. Simulations demonstrate that the proposed VDK-GP achieves more than two fold sample complexity reduction, compared to a generic GP on medium scale 500-Bus and large scale 1354-Bus power systems. Moreover, active learning achieves an impressive reduction of over 15 times in comparison to the time complexity of Monte-Carlo simulations (MCS), and have risk estimation error of order 1E-4 for both 500-Bus and 1354-Bus system, demonstrating its superior efficiency in risk estimation.
Authors: Yunho Kim, Hyunsik Oh, Jeonghyun Lee, Jinhyeok Choi, Gwanghyeon Ji, Moonkyu Jung, Donghoon Youm, Jemin Hwangbo
Abstract: Several earlier studies have shown impressive control performance in complex robotic systems by designing the controller using a neural network and training it with model-free reinforcement learning. However, these outstanding controllers with natural motion style and high task performance are developed through extensive reward engineering, which is a highly laborious and time-consuming process of designing numerous reward terms and determining suitable reward coefficients. In this work, we propose a novel reinforcement learning framework for training neural network controllers for complex robotic systems consisting of both rewards and constraints. To let the engineers appropriately reflect their intent to constraints and handle them with minimal computation overhead, two constraint types and an efficient policy optimization algorithm are suggested. The learning framework is applied to train locomotion controllers for several legged robots with different morphology and physical attributes to traverse challenging terrains. Extensive simulation and real-world experiments demonstrate that performant controllers can be trained with significantly less reward engineering, by tuning only a single reward coefficient. Furthermore, a more straightforward and intuitive engineering process can be utilized, thanks to the interpretability and generalizability of constraints. The summary video is available at https://youtu.be/KAlm3yskhvM.
Authors: Zhichao Huang, Chutong Meng, Tom Ko
Abstract: With recent rapid growth of large language models (LLMs), discrete speech tokenization has played an important role for injecting speech into LLMs. However, this discretization gives rise to a loss of information, consequently impairing overall performance. To improve the performance of these discrete speech tokens, we present RepCodec, a novel speech representation codec for semantic speech tokenization. In contrast to audio codecs which reconstruct the raw audio, RepCodec learns a vector quantization codebook through reconstructing speech representations from speech encoders like HuBERT or data2vec. Together, the speech encoder, the codec encoder and the vector quantization codebook form a pipeline for converting speech waveforms into semantic tokens. The extensive experiments illustrate that RepCodec, by virtue of its enhanced information retention capacity, significantly outperforms the widely used k-means clustering approach in both speech understanding and generation. Furthermore, this superiority extends across various speech encoders and languages, affirming the robustness of RepCodec. We believe our method can facilitate large language modeling research on speech processing.
Authors: Constantinos Daskalakis, Noah Golowich, Nika Haghtalab, Abhishek Shetty
Abstract: A fundamental shortcoming of the concept of Nash equilibrium is its computational intractability: approximating Nash equilibria in normal-form games is PPAD-hard. In this paper, inspired by the ideas of smoothed analysis, we introduce a relaxed variant of Nash equilibrium called $\sigma$-smooth Nash equilibrium, for a smoothness parameter $\sigma$. In a $\sigma$-smooth Nash equilibrium, players only need to achieve utility at least as high as their best deviation to a $\sigma$-smooth strategy, which is a distribution that does not put too much mass (as parametrized by $\sigma$) on any fixed action. We distinguish two variants of $\sigma$-smooth Nash equilibria: strong $\sigma$-smooth Nash equilibria, in which players are required to play $\sigma$-smooth strategies under equilibrium play, and weak $\sigma$-smooth Nash equilibria, where there is no such requirement. We show that both weak and strong $\sigma$-smooth Nash equilibria have superior computational properties to Nash equilibria: when $\sigma$ as well as an approximation parameter $\epsilon$ and the number of players are all constants, there is a constant-time randomized algorithm to find a weak $\epsilon$-approximate $\sigma$-smooth Nash equilibrium in normal-form games. In the same parameter regime, there is a polynomial-time deterministic algorithm to find a strong $\epsilon$-approximate $\sigma$-smooth Nash equilibrium in a normal-form game. These results stand in contrast to the optimal algorithm for computing $\epsilon$-approximate Nash equilibria, which cannot run in faster than quasipolynomial-time. We complement our upper bounds by showing that when either $\sigma$ or $\epsilon$ is an inverse polynomial, finding a weak $\epsilon$-approximate $\sigma$-smooth Nash equilibria becomes computationally intractable.
Authors: Eman Ali, Muhammad Haris Khan
Abstract: Recent advances in large-scale vision-language models have achieved impressive performance in various zero-shot image classification tasks. While prior studies have demonstrated significant improvements by introducing few-shot labelled target samples, they still require labelling of target samples, which greatly degrades their scalability and generalizability while handling various visual recognition tasks. We design NtUA, a Noise-tolerant Unsupervised Adapter that allows the learning of effective target models with few unlabelled target samples. NtUA works as a key-value cache that formulates visual features and predicted pseudo-labels of the few unlabelled target samples as key-value pairs. It consists of two complementary designs. The first is adaptive cache formation that combats pseudo-label noises by weighting the key-value pairs according to their prediction confidence. The second is knowledge-guided cache refinement, which refines pair values (i.e., pseudo-labels) and cache weights by leveraging knowledge distillation from large-scale vision language models. Extensive experiments show that NtUA achieves superior performance consistently across multiple widely adopted benchmarks.
Authors: Anil Batra, Davide Moltisanti, Laura Sevilla-Lara, Marcus Rohrbach, Frank Keller
Abstract: Procedural videos, exemplified by recipe demonstrations, are instrumental in conveying step-by-step instructions. However, understanding such videos is challenging as it involves the precise localization of steps and the generation of textual instructions. Manually annotating steps and writing instructions is costly, which limits the size of current datasets and hinders effective learning. Leveraging large but noisy video-transcript datasets for pre-training can boost performance but demands significant computational resources. Furthermore, transcripts contain irrelevant content and differ in style from human-written instructions. To mitigate these issues, we propose a novel technique, Sieve-&-Swap, to automatically generate high-quality training data for the recipe domain: (i) Sieve: filters irrelevant transcripts and (ii) Swap: acquires high-quality text by replacing transcripts with human-written instruction from a text-only recipe dataset. The resulting dataset is three orders of magnitude smaller than current web-scale datasets but enables efficient training of large-scale models. Alongside Sieve-&-Swap, we propose Procedure Transformer (ProcX), a model for end-to-end step localization and instruction generation for procedural videos. When pre-trained on our curated dataset, this model achieves state-of-the-art performance on YouCook2 and Tasty while using a fraction of the training data. We have released code and dataset.
Authors: Chengshu Li, Jacky Liang, Andy Zeng, Xinyun Chen, Karol Hausman, Dorsa Sadigh, Sergey Levine, Li Fei-Fei, Fei Xia, Brian Ichter
Abstract: Code provides a general syntactic structure to build complex programs and perform precise computations when paired with a code interpreter - we hypothesize that language models (LMs) can leverage code-writing to improve Chain of Thought reasoning not only for logic and arithmetic tasks, but also for semantic ones (and in particular, those that are a mix of both). For example, consider prompting an LM to write code that counts the number of times it detects sarcasm in an essay: the LM may struggle to write an implementation for "detect_sarcasm(string)" that can be executed by the interpreter (handling the edge cases would be insurmountable). However, LMs may still produce a valid solution if they not only write code, but also selectively "emulate" the interpreter by generating the expected output of "detect_sarcasm(string)". In this work, we propose Chain of Code (CoC), a simple yet surprisingly effective extension that improves LM code-driven reasoning. The key idea is to encourage LMs to format semantic sub-tasks in a program as flexible pseudocode that the interpreter can explicitly catch undefined behaviors and hand off to simulate with an LM (as an "LMulator"). Experiments demonstrate that Chain of Code outperforms Chain of Thought and other baselines across a variety of benchmarks; on BIG-Bench Hard, Chain of Code achieves 84%, a gain of 12% over Chain of Thought. In a nutshell, CoC broadens the scope of reasoning questions that LMs can answer by "thinking in code".
Authors: Philipp Reiser, Javier Enrique Aguilar, Anneli Guthke, Paul-Christian B\"urkner
Abstract: Surrogate models are statistical or conceptual approximations for more complex simulation models. In this context, it is crucial to propagate the uncertainty induced by limited simulation budget and surrogate approximation error to predictions, inference, and subsequent decision-relevant quantities. However, quantifying and then propagating the uncertainty of surrogates is usually limited to special analytic cases or is otherwise computationally very expensive. In this paper, we propose a framework enabling a scalable, Bayesian approach to surrogate modeling with thorough uncertainty quantification, propagation, and validation. Specifically, we present three methods for Bayesian inference with surrogate models given measurement data. This is a task where the propagation of surrogate uncertainty is especially relevant, because failing to account for it may lead to biased and/or overconfident estimates of the parameters of interest. We showcase our approach in three detailed case studies for linear and nonlinear real-world modeling scenarios. Uncertainty propagation in surrogate models enables more reliable and safe approximation of expensive simulators and will therefore be useful in various fields of applications.
Authors: Haoran Ye, Jiarui Wang, Helan Liang, Zhiguang Cao, Yong Li, Fanzhang Li
Abstract: The recent end-to-end neural solvers have shown promise for small-scale routing problems but suffered from limited real-time scaling-up performance. This paper proposes GLOP (Global and Local Optimization Policies), a unified hierarchical framework that efficiently scales toward large-scale routing problems. GLOP partitions large routing problems into Travelling Salesman Problems (TSPs) and TSPs into Shortest Hamiltonian Path Problems. For the first time, we hybridize non-autoregressive neural heuristics for coarse-grained problem partitions and autoregressive neural heuristics for fine-grained route constructions, leveraging the scalability of the former and the meticulousness of the latter. Experimental results show that GLOP achieves competitive and state-of-the-art real-time performance on large-scale routing problems, including TSP, ATSP, CVRP, and PCTSP.
Authors: Sahar Arshad, Seemab Latif, Ahmad Salman, Rabia Latif
Abstract: Financial forecasting plays an important role in making informed decisions for financial stakeholders, specifically in the stock exchange market. In a traditional setting, investors commonly rely on the equity research department for valuable reports on market insights and investment recommendations. The equity research department, however, faces challenges in effectuating decision-making do to the demanding cognitive effort required for analyzing the inherently volatile nature of market dynamics. Furthermore, financial forecasting systems employed by analysts pose potential risks in terms of interpretability and gaining the trust of all stakeholders. This paper presents an interpretable decision-making model leveraging the SHAP-based explainability technique to forecast investment recommendations. The proposed solution not only provides valuable insights into the factors that influence forecasted recommendations but also caters the investors of varying types, including those interested in daily and short-term investment opportunities. To ascertain the efficacy of the proposed model, a case study is devised that demonstrates a notable enhancement in investor's portfolio value, employing our trading strategies. The results highlight the significance of incorporating interpretability in forecasting models to boost stakeholders' confidence and foster transparency in the stock exchange domain.
Authors: Reza Fayyazi, Rozhina Taghdimi, Shanchieh Jay Yang
Abstract: Tactics, Techniques, and Procedures (TTPs) outline the methods attackers use to exploit vulnerabilities. The interpretation of TTPs in the MITRE ATT&CK framework can be challenging for cybersecurity practitioners due to presumed expertise and complex dependencies. Meanwhile, advancements with Large Language Models (LLMs) have led to recent surge in studies exploring its uses in cybersecurity operations. It is, however, unclear how LLMs can be used in an efficient and proper way to provide accurate responses for critical domains such as cybersecurity. This leads us to investigate how to better use two types of LLMs: small-scale encoder-only (e.g., RoBERTa) and larger decoder-only (e.g., GPT-3.5) LLMs to comprehend and summarize TTPs with the intended purposes (i.e., tactics) of a cyberattack procedure. This work studies and compares the uses of supervised fine-tuning (SFT) of encoder-only LLMs vs. Retrieval Augmented Generation (RAG) for decoder-only LLMs (without fine-tuning). Both SFT and RAG techniques presumably enhance the LLMs with relevant contexts for each cyberattack procedure. Our studies show decoder-only LLMs with RAG achieves better performance than encoder-only models with SFT, particularly when directly relevant context is extracted by RAG. The decoder-only results could suffer low `Precision' while achieving high `Recall'. Our findings further highlight a counter-intuitive observation that more generic prompts tend to yield better predictions of cyberattack tactics than those that are more specifically tailored.
Authors: Mi{\l}osz Kasak, Kamil Deja, Maja Karwowska, Monika Jakubowska, {\L}ukasz Graczykowski, Ma{\l}gorzata Janik
Abstract: In this work, we introduce a novel method for Particle Identification (PID) within the scope of the ALICE experiment at the Large Hadron Collider at CERN. Identifying products of ultrarelativisitc collisions delivered by the LHC is one of the crucial objectives of ALICE. Typically employed PID methods rely on hand-crafted selections, which compare experimental data to theoretical simulations. To improve the performance of the baseline methods, novel approaches use machine learning models that learn the proper assignment in a classification task. However, because of the various detection techniques used by different subdetectors, as well as the limited detector efficiency and acceptance, produced particles do not always yield signals in all of the ALICE components. This results in data with missing values. Machine learning techniques cannot be trained with such examples, so a significant part of the data is skipped during training. In this work, we propose the first method for PID that can be trained with all of the available data examples, including incomplete ones. Our approach improves the PID purity and efficiency of the selected sample for all investigated particle species.
Authors: He Jia
Abstract: We present Neural Quantile Estimation (NQE), a novel Simulation-Based Inference (SBI) method based on conditional quantile regression. NQE autoregressively learns individual one dimensional quantiles for each posterior dimension, conditioned on the data and previous posterior dimensions. Posterior samples are obtained by interpolating the predicted quantiles using monotonic cubic Hermite spline, with specific treatment for the tail behavior and multi-modal distributions. We introduce an alternative definition for the Bayesian credible region using the local Cumulative Density Function (CDF), offering substantially faster evaluation than the traditional Highest Posterior Density Region (HPDR). In case of limited simulation budget and/or known model misspecification, a post-processing calibration step can be integrated into NQE to ensure the unbiasedness of the posterior estimation with negligible additional computational cost. We demonstrate that NQE achieves state-of-the-art performance on a variety of benchmark problems.
Authors: Vladim\'ir Bo\v{z}a
Abstract: Pruning large language models (LLMs) is a challenging task due to their enormous size. The primary difficulty is fine-tuning the model after pruning, which is needed to recover the lost performance caused by dropping weights. Recent approaches have either ignored fine-tuning entirely, focusing on efficient pruning criteria, or attempted layer-wise weight updates, preserving the behavior of each layer. However, even layer-wise weight updates can be costly for LLMs, and previous works have resorted to various approximations. In our paper, we propose a fast and effective weight update algorithm for pruned layers based on the Alternating Direction Method of Multipliers (ADMM). We further extend it with a simple gradual pruning mask selection and achieve state-of-the-art pruning performance across a wide range of LLMs. Code is available at https://github.com/fmfi-compbio/admm-pruning.
Authors: William G. Powell, Hanbaek Lyu
Abstract: For obtaining optimal first-order convergence guarantee for stochastic optimization, it is necessary to use a recurrent data sampling algorithm that samples every data point with sufficient frequency. Most commonly used data sampling algorithms (e.g., i.i.d., MCMC, random reshuffling) are indeed recurrent under mild assumptions. In this work, we show that for a particular class of stochastic optimization algorithms, we do not need any other property (e.g., independence, exponential mixing, and reshuffling) than recurrence in data sampling algorithms to guarantee the optimal rate of first-order convergence. Namely, using regularized versions of Minimization by Incremental Surrogate Optimization (MISO), we show that for non-convex and possibly non-smooth objective functions, the expected optimality gap converges at an optimal rate $O(n^{-1/2})$ under general recurrent sampling schemes. Furthermore, the implied constant depends explicitly on the `speed of recurrence', measured by the expected amount of time to visit a given data point either averaged (`target time') or supremized (`hitting time') over the current location. We demonstrate theoretically and empirically that convergence can be accelerated by selecting sampling algorithms that cover the data set most effectively. We discuss applications of our general framework to decentralized optimization and distributed non-negative matrix factorization.
Authors: Zeliang Zhang, Wei Yao, Xiaosen Wang
Abstract: Deep neural networks are widely known to be vulnerable to adversarial examples. However, vanilla adversarial examples generated under the white-box setting often exhibit low transferability across different models. Since adversarial transferability poses more severe threats to practical applications, various approaches have been proposed for better transferability, including gradient-based, input transformation-based, and model-related attacks, \etc. In this work, we find that several tiny changes in the existing adversarial attacks can significantly affect the attack performance, \eg, the number of iterations and step size. Based on careful studies of existing adversarial attacks, we propose a bag of tricks to enhance adversarial transferability, including momentum initialization, scheduled step size, dual example, spectral-based input transformation, and several ensemble strategies. Extensive experiments on the ImageNet dataset validate the high effectiveness of our proposed tricks and show that combining them can further boost adversarial transferability. Our work provides practical insights and techniques to enhance adversarial transferability, and offers guidance to improve the attack performance on the real-world application through simple adjustments.
Authors: Nicholas Galioto, Harsh Sharma, Boris Kramer, Alex Arkady Gorodetsky
Abstract: This paper presents a structure-preserving Bayesian approach for learning nonseparable Hamiltonian systems using stochastic dynamic models allowing for statistically-dependent, vector-valued additive and multiplicative measurement noise. The approach is comprised of three main facets. First, we derive a Gaussian filter for a statistically-dependent, vector-valued, additive and multiplicative noise model that is needed to evaluate the likelihood within the Bayesian posterior. Second, we develop a novel algorithm for cost-effective application of Bayesian system identification to high-dimensional systems. Third, we demonstrate how structure-preserving methods can be incorporated into the proposed framework, using nonseparable Hamiltonians as an illustrative system class. We assess the method's performance based on the forecasting accuracy of a model estimated from single-trajectory data. We compare the Bayesian method to a state-of-the-art machine learning method on a canonical nonseparable Hamiltonian model and a chaotic double pendulum model with small, noisy training datasets. The results show that using the Bayesian posterior as a training objective can yield upwards of 724 times improvement in Hamiltonian mean squared error using training data with up to 10% multiplicative noise compared to a standard training objective. Lastly, we demonstrate the utility of the novel algorithm for parameter estimation of a 64-dimensional model of the spatially-discretized nonlinear Schr\"odinger equation with data corrupted by up to 20% multiplicative noise.
Authors: Bodo Rosenhahn, Christoph Hirche
Abstract: A Normalizing Flow computes a bijective mapping from an arbitrary distribution to a predefined (e.g. normal) distribution. Such a flow can be used to address different tasks, e.g. anomaly detection, once such a mapping has been learned. In this work we introduce Normalizing Flows for Quantum architectures, describe how to model and optimize such a flow and evaluate our method on example datasets. Our proposed models show competitive performance for anomaly detection compared to classical methods, esp. those ones where there are already quantum inspired algorithms available. In the experiments we compare our performance to isolation forests (IF), the local outlier factor (LOF) or single-class SVMs.
Authors: Amin Parchami-Araghi, Moritz B\"ohle, Sukrut Rao, Bernt Schiele
Abstract: Knowledge Distillation (KD) has proven effective for compressing large teacher models into smaller student models. While it is well known that student models can achieve similar accuracies as the teachers, it has also been shown that they nonetheless often do not learn the same function. It is, however, often highly desirable that the student's and teacher's functions share similar properties such as basing the prediction on the same input features, as this ensures that students learn the 'right features' from the teachers. In this work, we explore whether this can be achieved by not only optimizing the classic KD loss but also the similarity of the explanations generated by the teacher and the student. Despite the idea being simple and intuitive, we find that our proposed 'explanation-enhanced' KD (e$^2$KD) (1) consistently provides large gains in terms of accuracy and student-teacher agreement, (2) ensures that the student learns from the teacher to be right for the right reasons and to give similar explanations, and (3) is robust with respect to the model architectures, the amount of training data, and even works with 'approximate', pre-computed explanations.
Authors: Chenlu Ye, Jiafan He, Quanquan Gu, Tong Zhang
Abstract: This study tackles the challenges of adversarial corruption in model-based reinforcement learning (RL), where the transition dynamics can be corrupted by an adversary. Existing studies on corruption-robust RL mostly focus on the setting of model-free RL, where robust least-square regression is often employed for value function estimation. However, these techniques cannot be directly applied to model-based RL. In this paper, we focus on model-based RL and take the maximum likelihood estimation (MLE) approach to learn transition model. Our work encompasses both online and offline settings. In the online setting, we introduce an algorithm called corruption-robust optimistic MLE (CR-OMLE), which leverages total-variation (TV)-based information ratios as uncertainty weights for MLE. We prove that CR-OMLE achieves a regret of $\tilde{\mathcal{O}}(\sqrt{T} + C)$, where $C$ denotes the cumulative corruption level after $T$ episodes. We also prove a lower bound to show that the additive dependence on $C$ is optimal. We extend our weighting technique to the offline setting, and propose an algorithm named corruption-robust pessimistic MLE (CR-PMLE). Under a uniform coverage condition, CR-PMLE exhibits suboptimality worsened by $\mathcal{O}(C/n)$, nearly matching the lower bound. To the best of our knowledge, this is the first work on corruption-robust model-based RL algorithms with provable guarantees.
Authors: Zhijian Duan, Haoran Sun, Yichong Xia, Siqiang Wang, Zhilin Zhang, Chuan Yu, Jian Xu, Bo Zheng, Xiaotie Deng
Abstract: Identifying high-revenue mechanisms that are both dominant strategy incentive compatible (DSIC) and individually rational (IR) is a fundamental challenge in auction design. While theoretical approaches have encountered bottlenecks in multi-item auctions, there has been much empirical progress in automated designing such mechanisms using machine learning. However, existing research primarily focuses on randomized auctions, with less attention given to the more practical deterministic auctions. Therefore, this paper investigates the automated design of deterministic auctions and introduces OD-VVCA, an objective decomposition approach for automated designing Virtual Valuations Combinatorial Auctions (VVCAs). Firstly, we restrict our mechanism to deterministic VVCAs, which are inherently DSIC and IR. Afterward, we utilize a parallelizable dynamic programming algorithm to compute the allocation and revenue outcomes of a VVCA efficiently. We then decompose the revenue objective function into continuous and piecewise constant discontinuous components, optimizing each using distinct methods. Extensive experiments show that OD-VVCA achieves high revenue in multi-item auctions, especially in large-scale settings where it outperforms both randomized and deterministic baselines, indicating its efficacy and scalability.
Authors: Mikayel Samvelyan, Sharath Chandra Raparthy, Andrei Lupu, Eric Hambro, Aram H. Markosyan, Manish Bhatt, Yuning Mao, Minqi Jiang, Jack Parker-Holder, Jakob Foerster, Tim Rockt\"aschel, Roberta Raileanu
Abstract: As large language models (LLMs) become increasingly prevalent across many real-world applications, understanding and enhancing their robustness to adversarial attacks is of paramount importance. Existing methods for identifying adversarial prompts tend to focus on specific domains, lack diversity, or require extensive human annotations. To address these limitations, we present Rainbow Teaming, a novel black-box approach for producing a diverse collection of adversarial prompts. Rainbow Teaming casts adversarial prompt generation as a quality-diversity problem, and uses open-ended search to generate prompts that are both effective and diverse. Focusing on the safety domain, we use Rainbow Teaming to target various state-of-the-art LLMs, including the Llama 2 and Llama 3 models. Our approach reveals hundreds of effective adversarial prompts, with an attack success rate exceeding 90% across all tested models. Furthermore, we demonstrate that fine-tuning models with synthetic data generated by the Rainbow Teaming method significantly enhances their safety without sacrificing general performance or helpfulness. We additionally explore the versatility of Rainbow Teaming by applying it to question answering and cybersecurity, showcasing its potential to drive robust open-ended self-improvement in a wide range of applications.
Authors: Yunwei Bai, Ying Kiat Tan, Tsuhan Chen
Abstract: Few-shot-learning (FSL) commonly requires a model to identify images (queries) that belong to classes unseen during training, based on a few labelled samples of the new classes (support set) as reference. So far, plenty of algorithms involve training data augmentation to improve the generalization capability of FSL models, but outlier query or support images during inference can still pose great generalization challenges. In this work, to reduce the bias caused by unconventional test samples, we generate additional test-class samples through combining them with suitable train-class samples, whose averaged representations then become more typical for few-shot classification. We experimentally and theoretically demonstrate the effectiveness of our method, via which we can achieve around 4\% test accuracy improvement for trained FSL models. Importantly, our method is universally compatible with different off-the-shelf FSL models, whose performance can be improved without extra dataset nor further training of the models themselves.
Authors: Fahimeh Hosseini Noohdani, Parsa Hosseini, Aryan Yazdan Parast, Hamidreza Yaghoubi Araghi, Mahdieh Soleymani Baghshah
Abstract: While standard Empirical Risk Minimization (ERM) training is proven effective for image classification on in-distribution data, it fails to perform well on out-of-distribution samples. One of the main sources of distribution shift for image classification is the compositional nature of images. Specifically, in addition to the main object or component(s) determining the label, some other image components usually exist, which may lead to the shift of input distribution between train and test environments. More importantly, these components may have spurious correlations with the label. To address this issue, we propose Decompose-and-Compose (DaC), which improves robustness to correlation shift by a compositional approach based on combining elements of images. Based on our observations, models trained with ERM usually highly attend to either the causal components or the components having a high spurious correlation with the label (especially in datapoints on which models have a high confidence). In fact, according to the amount of spurious correlation and the easiness of classification based on the causal or non-causal components, the model usually attends to one of these more (on samples with high confidence). Following this, we first try to identify the causal components of images using class activation maps of models trained with ERM. Afterward, we intervene on images by combining them and retraining the model on the augmented data, including the counterfactual ones. Along with its high interpretability, this work proposes a group-balancing method by intervening on images without requiring group labels or information regarding the spurious features during training. The method has an overall better worst group accuracy compared to previous methods with the same amount of supervision on the group labels in correlation shift.
Authors: Daniel Fl\"ogel, Lars Fischer, Thomas Rudolf, Tobias Sch\"urmann, S\"oren Hohmann
Abstract: Mobile robots are being used on a large scale in various crowded situations and become part of our society. The socially acceptable navigation behavior of a mobile robot with individual human consideration is an essential requirement for scalable applications and human acceptance. Deep Reinforcement Learning (DRL) approaches are recently used to learn a robot's navigation policy and to model the complex interactions between robots and humans. We propose to divide existing DRL-based navigation approaches based on the robot's exhibited social behavior and distinguish between social collision avoidance with a lack of social behavior and socially aware approaches with explicit predefined social behavior. In addition, we propose a novel socially integrated navigation approach where the robot's social behavior is adaptive and emerges from the interaction with humans. The formulation of our approach is derived from a sociological definition, which states that social acting is oriented toward the acting of others. The DRL policy is trained in an environment where other agents interact socially integrated and reward the robot's behavior individually. The simulation results indicate that the proposed socially integrated navigation approach outperforms a socially aware approach in terms of ego navigation performance while significantly reducing the negative impact on all agents within the environment.
Authors: Siavash Jafarzadeh, Stewart Silling, Lu Zhang, Colton Ross, Chung-Hao Lee, S. M. Rakibur Rahman, Shuodao Wang, Yue Yu
Abstract: Human tissues are highly organized structures with collagen fiber arrangements varying from point to point. Anisotropy of the tissue arises from the natural orientation of the fibers, resulting in location-dependent anisotropy. Heterogeneity also plays an important role in tissue function. It is therefore critical to discover and understand the distribution of fiber orientations from experimental mechanical measurements such as digital image correlation (DIC) data. To this end, we introduce the Heterogeneous Peridynamic Neural Operator (HeteroPNO) approach for data-driven constitutive modeling of heterogeneous anisotropic materials. Our goal is to learn a nonlocal constitutive law together with the material microstructure, in the form of a heterogeneous fiber orientation field, from load-displacement field measurements. We propose a two-phase learning approach. Firstly, we learn a homogeneous constitutive law in the form of a neural network-based kernel function and a nonlocal bond force, to capture complex homogeneous material responses from data. Then, in the second phase we reinitialize the learnt bond force and the kernel function, and training them together with a fiber orientation field for each material point. Owing to the state-based peridynamic skeleton, our HeteroPNO-learned material models are objective and have the balance of linear and angular momentum guaranteed. Moreover, the effects from heterogeneity and nonlinear constitutive relationship are captured by the kernel function and the bond force respectively, enabling physical interpretability. As a result, our HeteroPNO architecture can learn a constitutive model for a biological tissue with anisotropic heterogeneous response undergoing large deformation regime. Moreover, the framework is capable to provide displacement and stress field predictions for new and unseen loading instances.
Authors: Thibaut Thonet, Jos Rozen, Laurent Besacier
Abstract: Research on Large Language Models (LLMs) has recently witnessed an increasing interest in extending models' context size to better capture dependencies within long documents. While benchmarks have been proposed to assess long-range abilities, existing efforts primarily considered generic tasks that are not necessarily aligned with real-world applications. In contrast, our work proposes a new benchmark for long-context LLMs focused on a practical meeting assistant scenario. In this scenario, the long contexts consist of transcripts obtained by automatic speech recognition, presenting unique challenges for LLMs due to the inherent noisiness and oral nature of such data. Our benchmark, named ELITR-Bench, augments the existing ELITR corpus' transcripts with 271 manually crafted questions and their ground-truth answers. Our experiments with recent long-context LLMs on ELITR-Bench highlight a gap between open-source and proprietary models, especially when questions are asked sequentially within a conversation. We also provide a thorough analysis of our GPT-4-based evaluation method, encompassing insights from a crowdsourcing study. Our findings suggest that while GPT-4's evaluation scores are correlated with human judges', its ability to differentiate among more than three score levels may be limited.
Authors: Shiming Chen, Wenjin Hou, Salman Khan, Fahad Shahbaz Khan
Abstract: Zero-shot learning (ZSL) recognizes the unseen classes by conducting visual-semantic interactions to transfer semantic knowledge from seen classes to unseen ones, supported by semantic information (e.g., attributes). However, existing ZSL methods simply extract visual features using a pre-trained network backbone (i.e., CNN or ViT), which fail to learn matched visual-semantic correspondences for representing semantic-related visual features as lacking of the guidance of semantic information, resulting in undesirable visual-semantic interactions. To tackle this issue, we propose a progressive semantic-guided vision transformer for zero-shot learning (dubbed ZSLViT). ZSLViT mainly considers two properties in the whole network: i) discover the semantic-related visual representations explicitly, and ii) discard the semantic-unrelated visual information. Specifically, we first introduce semantic-embedded token learning to improve the visual-semantic correspondences via semantic enhancement and discover the semantic-related visual tokens explicitly with semantic-guided token attention. Then, we fuse low semantic-visual correspondence visual tokens to discard the semantic-unrelated visual information for visual enhancement. These two operations are integrated into various encoders to progressively learn semantic-related visual representations for accurate visual-semantic interactions in ZSL. The extensive experiments show that our ZSLViT achieves significant performance gains on three popular benchmark datasets, i.e., CUB, SUN, and AWA2. Codes are available at: https://github.com/shiming-chen/ZSLViT .
Authors: Ming Li, Taojiannan Yang, Huafeng Kuang, Jie Wu, Zhaoning Wang, Xuefeng Xiao, Chen Chen
Abstract: To enhance the controllability of text-to-image diffusion models, existing efforts like ControlNet incorporated image-based conditional controls. In this paper, we reveal that existing methods still face significant challenges in generating images that align with the image conditional controls. To this end, we propose ControlNet++, a novel approach that improves controllable generation by explicitly optimizing pixel-level cycle consistency between generated images and conditional controls. Specifically, for an input conditional control, we use a pre-trained discriminative reward model to extract the corresponding condition of the generated images, and then optimize the consistency loss between the input conditional control and extracted condition. A straightforward implementation would be generating images from random noises and then calculating the consistency loss, but such an approach requires storing gradients for multiple sampling timesteps, leading to considerable time and memory costs. To address this, we introduce an efficient reward strategy that deliberately disturbs the input images by adding noise, and then uses the single-step denoised images for reward fine-tuning. This avoids the extensive costs associated with image sampling, allowing for more efficient reward fine-tuning. Extensive experiments show that ControlNet++ significantly improves controllability under various conditional controls. For example, it achieves improvements over ControlNet by 11.1% mIoU, 13.4% SSIM, and 7.6% RMSE, respectively, for segmentation mask, line-art edge, and depth conditions. All the code, models, demo and organized data have been open sourced on our Github Repo.
Authors: Ahmad Halimi Razlighi, Carsten Bockelmann, Armin Dekorsy
Abstract: In this paper, we investigated semantic communication for multi-task processing using an information-theoretic approach. We introduced the concept of a "semantic source", allowing multiple semantic interpretations from a single observation. We formulated an end-to-end optimization problem taking into account the communication channel, maximizing mutual information (infomax) to design the semantic encoding and decoding process exploiting the statistical relations between semantic variables. To solve the problem we perform data-driven deep learning employing variational approximation techniques. Our semantic encoder is divided into a common unit and multiple specific units to facilitate cooperative multi-task processing. Simulation results demonstrate the effectiveness of our proposed semantic source and system design when statistical relationships exist, comparing cooperative task processing with independent task processing. However, our findings highlight that cooperative multi-tasking is not always beneficial, emphasizing the importance of statistical relationships between tasks and indicating the need for further investigation into the semantically processing of multiple tasks.
Authors: Giannis Daras, Alexandros G. Dimakis, Constantinos Daskalakis
Abstract: Ambient diffusion is a recently proposed framework for training diffusion models using corrupted data. Both Ambient Diffusion and alternative SURE-based approaches for learning diffusion models from corrupted data resort to approximations which deteriorate performance. We present the first framework for training diffusion models that provably sample from the uncorrupted distribution given only noisy training data, solving an open problem in this space. Our key technical contribution is a method that uses a double application of Tweedie's formula and a consistency loss function that allows us to extend sampling at noise levels below the observed data noise. We also provide further evidence that diffusion models memorize from their training sets by identifying extremely corrupted images that are almost perfectly reconstructed, raising copyright and privacy concerns. Our method for training using corrupted samples can be used to mitigate this problem. We demonstrate this by fine-tuning Stable Diffusion XL to generate samples from a distribution using only noisy samples. Our framework reduces the amount of memorization of the fine-tuning dataset, while maintaining competitive performance.
Authors: Xiang Li, Shunpan Liang, Yu Lei, Chen Li, Yulei Hou, Tengfei Ma
Abstract: Medication recommendation systems are developed to recommend suitable medications tailored to specific patient. Previous researches primarily focus on learning medication representations, which have yielded notable advances. However, these methods are limited to capturing personalized patient representations due to the following primary limitations: (i) unable to capture the differences in the impact of diseases/procedures on patients across various patient health states; (ii) fail to model the direct causal relationships between medications and specific health state of patients, resulting in an inability to determine which specific disease each medication is treating. To address these limitations, we propose CausalMed, a patient health state-centric model capable of enhancing the personalization of patient representations. Specifically, CausalMed first captures the causal relationship between diseases/procedures and medications through causal discovery and evaluates their causal effects. Building upon this, CausalMed focuses on analyzing the health state of patients, capturing the dynamic differences of diseases/procedures in different health states of patients, and transforming diseases/procedures into medications on direct causal relationships. Ultimately, CausalMed integrates information from longitudinal visits to recommend medication combinations. Extensive experiments on real-world datasets show that our method learns more personalized patient representation and outperforms state-of-the-art models in accuracy and safety.
Authors: Tyler Griggs, Xiaoxuan Liu, Jiaxiang Yu, Doyoung Kim, Wei-Lin Chiang, Alvin Cheung, Ion Stoica
Abstract: Large language models (LLMs) are increasingly integrated into many online services, yet they remain cost-prohibitive to deploy due to the requirement of expensive GPU instances. Prior work has addressed the high cost of LLM serving by improving the inference engine, but less attention has been given to selecting the most cost-efficient GPU type(s) for a specific LLM service. There is a large and growing landscape of GPU types and, within these options, higher cost does not always lead to increased performance. Instead, through a comprehensive investigation, we find that three key LLM service characteristics (request size, request rate, SLO) strongly influence GPU cost efficiency, and differing GPU types are most cost efficient for differing LLM service settings. As a result, the most cost-efficient allocation for a given service is typically a mix of heterogeneous GPU types. Based on this analysis, we introduce M\'elange, a GPU allocation framework that navigates these diverse LLM service characteristics and heterogeneous GPU option space to automatically and efficiently derive the minimal-cost GPU allocation for a given LLM service. We formulate the GPU allocation task as a cost-aware bin packing problem where GPUs are bins and items are slices of the service workload. Our formulation's constraints account for a service's unique characteristics, allowing M\'elange to be flexible to support diverse service settings and heterogeneity-aware to adapt the GPU allocation to a specific service. Compared to using only a single GPU type, M\'elange reduces deployment costs by up to 77% in conversational settings, 33% in document-based settings, and 51% in a mixed setting.
Authors: Ling Yue, Sixue Xing, Jintai Chen, Tianfan Fu
Abstract: Large Language Models (LLMs) and multi-agent systems have shown impressive capabilities in natural language tasks but face challenges in clinical trial applications, primarily due to limited access to external knowledge. Recognizing the potential of advanced clinical trial tools that aggregate and predict based on the latest medical data, we propose an integrated solution to enhance their accessibility and utility. We introduce Clinical Agent System (ClinicalAgent), a clinical multi-agent system designed for clinical trial tasks, leveraging GPT-4, multi-agent architectures, LEAST-TO-MOST, and ReAct reasoning technology. This integration not only boosts LLM performance in clinical contexts but also introduces novel functionalities. The proposed method achieves competitive predictive performance in clinical trial outcome prediction (0.7908 PR-AUC), obtaining a 0.3326 improvement over the standard prompt Method. Publicly available code can be found at https://anonymous.4open.science/r/ClinicalAgent-6671.
Authors: Henry Peng Zou, Vinay Samuel, Yue Zhou, Weizhi Zhang, Liancheng Fang, Zihe Song, Philip S. Yu, Cornelia Caragea
Abstract: Existing datasets for attribute value extraction (AVE) predominantly focus on explicit attribute values while neglecting the implicit ones, lack product images, are often not publicly available, and lack an in-depth human inspection across diverse domains. To address these limitations, we present ImplicitAVE, the first, publicly available multimodal dataset for implicit attribute value extraction. ImplicitAVE, sourced from the MAVE dataset, is carefully curated and expanded to include implicit AVE and multimodality, resulting in a refined dataset of 68k training and 1.6k testing data across five domains. We also explore the application of multimodal large language models (MLLMs) to implicit AVE, establishing a comprehensive benchmark for MLLMs on the ImplicitAVE dataset. Six recent MLLMs with eleven variants are evaluated across diverse settings, revealing that implicit value extraction remains a challenging task for MLLMs. The contributions of this work include the development and release of ImplicitAVE, and the exploration and benchmarking of various MLLMs for implicit AVE, providing valuable insights and potential future research directions. Dataset and code are available at https://github.com/HenryPengZou/ImplicitAVE
Authors: Zhenrong Zhang, Jianan Liu, Xi Zhou, Tao Huang, Qing-Long Han, Jingxin Liu, Hongbin Liu
Abstract: Cooperative perception is essential to enhance the efficiency and safety of future transportation systems, requiring extensive data sharing among vehicles on the road, which raises significant privacy concerns. Federated learning offers a promising solution by enabling data privacy-preserving collaborative enhancements in perception, decision-making, and planning among connected and autonomous vehicles (CAVs). However, federated learning is impeded by significant challenges arising from data heterogeneity across diverse clients, potentially diminishing model accuracy and prolonging convergence periods. This study introduces a specialized federated learning framework for CP, termed the federated dynamic weighted aggregation (FedDWA) algorithm, facilitated by dynamic adjusting loss (DALoss) function. This framework employs dynamic client weighting to direct model convergence and integrates a novel loss function that utilizes Kullback-Leibler divergence (KLD) to counteract the detrimental effects of non-independently and identically distributed (Non-IID) and unbalanced data. Utilizing the BEV transformer as the primary model, our rigorous testing on the OpenV2V dataset, augmented with FedBEVT data, demonstrates significant improvements in the average intersection over union (IoU). These results highlight the substantial potential of our federated learning framework to address data heterogeneity challenges in CP, thereby enhancing the accuracy of environmental perception models and facilitating more robust and efficient collaborative learning solutions in the transportation sector.
Authors: Yanjun Fu, Ethan Baker, Yu Ding, Yizheng Chen
Abstract: Code Large Language Models (Code LLMs) have been increasingly used by developers to boost productivity, but they often generate vulnerable code. Thus, there is an urgent need to ensure that code generated by Code LLMs is correct and secure. Previous research has primarily focused on generating secure code, overlooking the fact that secure code also needs to be correct. This oversight can lead to a false sense of security. Currently, the community lacks a method to measure actual progress in this area, and we need solutions that address both security and correctness of code generation. This paper introduces a new benchmark, CodeGuard+, along with two new metrics, to measure Code LLMs' ability to generate both secure and correct code. Using our new evaluation methods, we show that the state-of-the-art defense technique, prefix tuning, may not be as strong as previously believed, since it generates secure code but sacrifices functional correctness. We also demonstrate that different decoding methods significantly affect the security of Code LLMs. Furthermore, we explore a new defense direction: constrained decoding for secure code generation. We propose new constrained decoding techniques to generate secure code. Our results reveal that constrained decoding is more effective than prefix tuning to improve the security of Code LLMs, without requiring a specialized training dataset. Moreover, our evaluations over eight state-of-the-art Code LLMs show that constrained decoding has strong performance to improve the security of Code LLMs, and our technique outperforms GPT-4.
Authors: Shun Takagi, Li Xiong, Fumiyuki Kato, Yang Cao, Masatoshi Yoshikawa
Abstract: Human mobility data offers valuable insights for many applications such as urban planning and pandemic response, but its use also raises privacy concerns. In this paper, we introduce the Hierarchical and Multi-Resolution Network (HRNet), a novel deep generative model specifically designed to synthesize realistic human mobility data while guaranteeing differential privacy. We first identify the key difficulties inherent in learning human mobility data under differential privacy. In response to these challenges, HRNet integrates three components: a hierarchical location encoding mechanism, multi-task learning across multiple resolutions, and private pre-training. These elements collectively enhance the model's ability under the constraints of differential privacy. Through extensive comparative experiments utilizing a real-world dataset, HRNet demonstrates a marked improvement over existing methods in balancing the utility-privacy trade-off.
Authors: \'Alvaro Becerra, Javier Irigoyen, Roberto Daza, Ruth Cobos, Aythami Morales, Julian Fierrez, Mutlu Cukurova
Abstract: In this article, we explore computer vision approaches to detect abnormal head pose during e-learning sessions and we introduce a study on the effects of mobile phone usage during these sessions. We utilize behavioral data collected from 120 learners monitored while participating in a MOOC learning sessions. Our study focuses on the influence of phone-usage events on behavior and physiological responses, specifically attention, heart rate, and meditation, before, during, and after phone usage. Additionally, we propose an approach for estimating head pose events using images taken by the webcam during the MOOC learning sessions to detect phone-usage events. Our hypothesis suggests that head posture undergoes significant changes when learners interact with a mobile phone, contrasting with the typical behavior seen when learners face a computer during e-learning sessions. We propose an approach designed to detect deviations in head posture from the average observed during a learner's session, operating as a semi-supervised method. This system flags events indicating alterations in head posture for subsequent human review and selection of mobile phone usage occurrences with a sensitivity over 90%.
Authors: Jan Melechovsky, Abhinaba Roy, Dorien Herremans
Abstract: Generative models guided by text prompts are increasingly becoming more popular. However, no text-to-MIDI models currently exist due to the lack of a captioned MIDI dataset. This work aims to enable research that combines LLMs with symbolic music by presenting, the first openly available large-scale MIDI dataset with text captions. MIDI (Musical Instrument Digital Interface) files are widely used for encoding musical information and can capture the nuances of musical composition. They are widely used by music producers, composers, musicologists, and performers alike. Inspired by recent advancements in captioning techniques, we present a curated dataset of over 168k MIDI files with textual descriptions. Each MIDI caption describes the musical content, including tempo, chord progression, time signature, instruments, genre, and mood, thus facilitating multi-modal exploration and analysis. The dataset encompasses various genres, styles, and complexities, offering a rich data source for training and evaluating models for tasks such as music information retrieval, music understanding, and cross-modal translation. We provide detailed statistics about the dataset and have assessed the quality of the captions in an extensive listening study. We anticipate that this resource will stimulate further research at the intersection of music and natural language processing, fostering advancements in both fields.
Authors: Samar Agnihotri, Renu Rameshan, Ritwik Ghosal
Abstract: Lossless image compression is required in various applications to reduce storage or transmission costs of images, while requiring the reconstructed images to have zero information loss compared to the original. Existing lossless image compression methods either have simple design but poor compression performance, or complex design, better performance, but with no performance guarantees. In our endeavor to develop a lossless image compression method with low complexity and guaranteed performance, we argue that compressibility of a color image is essentially derived from the patterns in its spatial structure, intensity variations, and color variations. Thus, we divide the overall design of a lossless image compression scheme into three parts that exploit corresponding redundancies. We further argue that the binarized version of an image captures its fundamental spatial structure. In this first part of our work, we propose a scheme for lossless compression of binary images. The proposed scheme first learns dictionaries of $16\times16$, $8\times8$, $4\times4$, and $2\times 2$ square pixel patterns from various datasets of binary images. It then uses these dictionaries to encode binary images. These dictionaries have various interesting properties that are further exploited to construct an efficient and scalable scheme. Our preliminary results show that the proposed scheme consistently outperforms existing conventional and learning based lossless compression approaches, and provides, on average, as much as $1.5\times$ better performance than a common general purpose lossless compression scheme (WebP), more than $3\times$ better performance than a state of the art learning based scheme, and better performance than a specialized scheme for binary image compression (JBIG2).
Authors: Huma Ameer, Seemab Latif, Iram Tariq Bhatti, Rabia Latif
Abstract: The automated classification of stuttered speech has significant implications for timely assessments providing assistance to speech language pathologists. Despite notable advancements in the field, the cases in which multiple disfluencies occur in speech require attention. We have taken a progressive approach to fill this gap by classifying multi-stuttered speech more efficiently. The problem has been addressed by firstly curating a dataset of multi-stuttered disfluencies from open source dataset SEP-28k audio clips. Secondly, employing Whisper, a state-of-the-art speech recognition model has been leveraged by using its encoder and taking the problem as multi label classification. Thirdly, using a 6 encoder layer Whisper and experimenting with various layer freezing strategies, a computationally efficient configuration of the model was identified. The proposed configuration achieved micro, macro, and weighted F1-scores of 0.88, 0.85, and 0.87, correspondingly on an external test dataset i.e. Fluency-Bank. In addition, through layer freezing strategies, we were able to achieve the aforementioned results by fine-tuning a single encoder layer, consequently, reducing the model's trainable parameters from 20.27 million to 3.29 million. This research study unveils the contribution of the last encoder layer in the identification of disfluencies in stuttered speech. Consequently, it has led to a computationally efficient approach, 83.7% less parameters to train, making the proposed approach more adaptable for various dialects and languages.
Authors: Felix Wagner, Wentian Xu, Pramit Saha, Ziyun Liang, Daniel Whitehouse, David Menon, Virginia Newcombe, Natalie Voets, J. Alison Noble, Konstantinos Kamnitsas
Abstract: Segmentation models for brain lesions in MRI are commonly developed for a specific disease and trained on data with a predefined set of MRI modalities. Each such model cannot segment the disease using data with a different set of MRI modalities, nor can it segment any other type of disease. Moreover, this training paradigm does not allow a model to benefit from learning from heterogeneous databases that may contain scans and segmentation labels for different types of brain pathologies and diverse sets of MRI modalities. Additionally, the sensitivity of patient data often prevents centrally aggregating data, necessitating a decentralized approach. Is it feasible to use Federated Learning (FL) to train a single model on client databases that contain scans and labels of different brain pathologies and diverse sets of MRI modalities? We demonstrate promising results by combining appropriate, simple, and practical modifications to the model and training strategy: Designing a model with input channels that cover the whole set of modalities available across clients, training with random modality drop, and exploring the effects of feature normalization methods. Evaluation on 7 brain MRI databases with 5 different diseases shows that such FL framework can train a single model that is shown to be very promising in segmenting all disease types seen during training. Importantly, it is able to segment these diseases in new databases that contain sets of modalities different from those in training clients. These results demonstrate, for the first time, the feasibility and effectiveness of using Federated Learning to train a single 3D segmentation model on decentralised data with diverse brain diseases and MRI modalities, a necessary step towards leveraging heterogeneous real-world databases. Code will be made available at: https://github.com/FelixWag/FL-MultiDisease-MRI
Authors: Khai Le-Duc, David Thulke, Hung-Phong Tran, Long Vo-Dang, Khai-Nguyen Nguyen, Truong-Son Hy, Ralf Schl\"uter
Abstract: Spoken Named Entity Recognition (NER) aims to extracting named entities from speech and categorizing them into types like person, location, organization, etc. In this work, we present VietMed-NER - the first spoken NER dataset in the medical domain. To our best knowledge, our real-world dataset is the largest spoken NER dataset in the world in terms of the number of entity types, featuring 18 distinct types. Secondly, we present baseline results using various state-of-the-art pre-trained models: encoder-only and sequence-to-sequence. We found that pre-trained multilingual models XLM-R outperformed all monolingual models on both reference text and ASR output. Also in general, encoders perform better than sequence-to-sequence models for the NER task. By simply translating, the transcript is applicable not just to Vietnamese but to other languages as well. All code, data and models are made publicly available here: https://github.com/leduckhai/MultiMed
Authors: Cheng Zhang
Abstract: Point forecasting in univariate random walks is an important but challenging research topic that has attracted numerous researchers. Unfortunately, traditional regression methods for this task often fail to surpass naive benchmarks due to data unpredictability. From a decision fusion perspective, this study proposes a novel forecasting method, which is derived from a variant definition of random walks, where the random error term for the future value is expressed as a positive random error multiplied by a direction sign. This method, based on the fusion of movement and naive predictions, does not require a loss function for optimization and can be optimized by estimating movement prediction accuracy on the validation set. This characteristic prevents the fusion method from reverting to traditional regression methods and allows it to integrate various machine learning and deep learning models for movement prediction. The method's efficacy is demonstrated through simulations and real-world data experiments. It reliably outperforms naive forecasts with moderate movement prediction accuracies, such as 0.55, and is superior to baseline models such as the ARIMA, linear regression, MLP, and LSTM networks in forecasting the S&P 500 index and Bitcoin prices. This method is particularly advantageous when accurate point predictions are challenging but accurate movement predictions are attainable, translating movement predictions into point forecasts in random walk contexts.
Authors: Akash Kundu, Aritra Sarkar, Abhishek Sadhu
Abstract: Quantum architecture search (QAS) is a promising direction for optimization and automated design of quantum circuits towards quantum advantage. Recent techniques in QAS focus on machine learning-based approaches from reinforcement learning, like deep Q-network. While multi-layer perceptron-based deep Q-networks have been applied for QAS, their interpretability remains challenging due to the high number of parameters. In this work, we evaluate the practicality of Kolmogorov-Arnold Networks (KANs) in QAS problems, analyzing their efficiency in the task of quantum state preparation and quantum chemistry. In quantum state preparation, our results show that in a noiseless scenario, the probability of success and the number of optimal quantum circuit configurations to generate the multi-qubit maximally entangled states are $2\times$ to $5\times$ higher than Multi-Layer perceptions (MLPs). Moreover, in noisy scenarios, KAN can achieve a better fidelity in approximating maximally entangled state than MLPs, where the performance of the MLP significantly depends on the choice of activation function. In tackling quantum chemistry problems, we enhance the recently proposed QAS algorithm by integrating Curriculum Reinforcement Learning (CRL) with a KAN structure instead of the traditional MLP. This modification allows us to design a parameterized quantum circuit that contains fewer 2-qubit gates and has a shallower depth, thereby improving the efficiency of finding the ground state of a chemical Hamiltonian. Further investigation reveals that KAN requires a significantly smaller number of learnable parameters compared to MLPs; however, the average time of executing each episode for KAN is higher.
Authors: Xiaolong He, April Tran, David M. Bortz, Youngsoo Choi
Abstract: The parametric greedy latent space dynamics identification (gLaSDI) framework has demonstrated promising potential for accurate and efficient modeling of high-dimensional nonlinear physical systems. However, it remains challenging to handle noisy data. To enhance robustness against noise, we incorporate the weak-form estimation of nonlinear dynamics (WENDy) into gLaSDI. In the proposed weak-form gLaSDI (WgLaSDI) framework, an autoencoder and WENDy are trained simultaneously to discover intrinsic nonlinear latent-space dynamics of high-dimensional data. Compared to the standard sparse identification of nonlinear dynamics (SINDy) employed in gLaSDI, WENDy enables variance reduction and robust latent space discovery, therefore leading to more accurate and efficient reduced-order modeling. Furthermore, the greedy physics-informed active learning in WgLaSDI enables adaptive sampling of optimal training data on the fly for enhanced modeling accuracy. The effectiveness of the proposed framework is demonstrated by modeling various nonlinear dynamical problems, including viscous and inviscid Burgers' equations, time-dependent radial advection, and the Vlasov equation for plasma physics. With data that contains 5-10% Gaussian white noise, WgLaSDI outperforms gLaSDI by orders of magnitude, achieving 1-7% relative errors. Compared with the high-fidelity models, WgLaSDI achieves 121 to 1,779x speed-up.
Authors: Diksha Sharma, Vivek Balasaheb Sabale, Parvinder Singh, Atul Kumar
Abstract: In the present study, we use cross-domain classification using quantum machine learning for quantum advantages to readdress the entanglement versus separability paradigm. The inherent structure of quantum states and its relation to a particular class of quantum states are used to intuitively classify testing states from domains different from training states, called \textit{cross-domain classification}. Using our quantum machine learning algorithm, we demonstrate efficient classifications of two-qubit mixed states into entangled and separable classes. For analyzing the quantumness of correlations, our model adequately classifies Bell diagonal states as zero and non-zero discord states. In addition, we also extend our analysis to evaluate the robustness of our model using random local unitary transformations. Our results demonstrate the potential of the quantum support vector machine for classifying quantum states across the multi-dimensional Hilbert space in comparison to classical support vector machines and neural networks.
Authors: Kotaro Ikeda, Tomoya Uda, Daisuke Okanohara, Sosuke Ito
Abstract: We discuss a connection between a generative model, called the diffusion model, and nonequilibrium thermodynamics for the Fokker-Planck equation, called stochastic thermodynamics. Based on the techniques of stochastic thermodynamics, we derive the speed-accuracy trade-off for the diffusion models, which is a trade-off relationship between the speed and accuracy of data generation in diffusion models. Our result implies that the entropy production rate in the forward process affects the errors in data generation. From a stochastic thermodynamic perspective, our results provide quantitative insight into how best to generate data in diffusion models. The optimal learning protocol is introduced by the conservative force in stochastic thermodynamics and the geodesic of space by the 2-Wasserstein distance in optimal transport theory. We numerically illustrate the validity of the speed-accuracy trade-off for the diffusion models with different noise schedules such as the cosine schedule, the conditional optimal transport, and the optimal transport.
Authors: Ananthu Aniraj, Cassio F. Dantas, Dino Ienco, Diego Marcos
Abstract: Computer vision methods that explicitly detect object parts and reason on them are a step towards inherently interpretable models. Existing approaches that perform part discovery driven by a fine-grained classification task make very restrictive assumptions on the geometric properties of the discovered parts; they should be small and compact. Although this prior is useful in some cases, in this paper we show that pre-trained transformer-based vision models, such as self-supervised DINOv2 ViT, enable the relaxation of these constraints. In particular, we find that a total variation (TV) prior, which allows for multiple connected components of any size, substantially outperforms previous work. We test our approach on three fine-grained classification benchmarks: CUB, PartImageNet and Oxford Flowers, and compare our results to previously published methods as well as a re-implementation of the state-of-the-art method PDiscoNet with a transformer-based backbone. We consistently obtain substantial improvements across the board, both on part discovery metrics and the downstream classification task, showing that the strong inductive biases in self-supervised ViT models require to rethink the geometric priors that can be used for unsupervised part discovery.
Authors: Yang Liu, Weixing Chen, Yongjie Bai, Guanbin Li, Wen Gao, Liang Lin
Abstract: Embodied Artificial Intelligence (Embodied AI) is crucial for achieving Artificial General Intelligence (AGI) and serves as a foundation for various applications that bridge cyberspace and the physical world. Recently, the emergence of Multi-modal Large Models (MLMs) and World Models (WMs) have attracted significant attention due to their remarkable perception, interaction, and reasoning capabilities, making them a promising architecture for the brain of embodied agents. However, there is no comprehensive survey for Embodied AI in the era of MLMs. In this survey, we give a comprehensive exploration of the latest advancements in Embodied AI. Our analysis firstly navigates through the forefront of representative works of embodied robots and simulators, to fully understand the research focuses and their limitations. Then, we analyze four main research targets: 1) embodied perception, 2) embodied interaction, 3) embodied agent, and 4) sim-to-real adaptation, covering the state-of-the-art methods, essential paradigms, and comprehensive datasets. Additionally, we explore the complexities of MLMs in virtual and real embodied agents, highlighting their significance in facilitating interactions in dynamic digital and physical environments. Finally, we summarize the challenges and limitations of embodied AI and discuss their potential future directions. We hope this survey will serve as a foundational reference for the research community and inspire continued innovation. The associated project can be found at https://github.com/HCPLab-SYSU/Embodied_AI_Paper_List.
URLs: https://github.com/HCPLab-SYSU/Embodied_AI_Paper_List.
Authors: Haiqian Yang, Anh Q. Nguyen, Dapeng Bi, Markus J. Buehler, Ming Guo
Abstract: During developmental processes such as embryogenesis, how a group of cells fold into specific structures, is a central question in biology that defines how living organisms form. Establishing tissue-level morphology critically relies on how every single cell decides to position itself relative to its neighboring cells. Despite its importance, it remains a major challenge to understand and predict the behavior of every cell within the living tissue over time during such intricate processes. To tackle this question, we propose a geometric deep learning model that can predict multicellular folding and embryogenesis, accurately capturing the highly convoluted spatial interactions among cells. We demonstrate that multicellular data can be represented with both granular and foam-like physical pictures through a unified graph data structure, considering both cellular interactions and cell junction networks. We successfully use our model to achieve two important tasks, interpretable 4-D morphological sequence alignment, and predicting local cell rearrangements before they occur at single-cell resolution. Furthermore, using an activation map and ablation studies, we demonstrate that cell geometries and cell junction networks together regulate local cell rearrangement which is critical for embryo morphogenesis. This approach provides a novel paradigm to study morphogenesis, highlighting a unified data structure and harnessing the power of geometric deep learning to accurately model the mechanisms and behaviors of cells during development. It offers a pathway toward creating a unified dynamic morphological atlas for a variety of developmental processes such as embryogenesis.
Authors: Md Ruman Islam (University of Nebraska Omaha), Raja Hasnain Anwar (University of Massachusetts Amherst), Spyridon Mastorakis (University of Notre Dame), Muhammad Taqi Raza (University of Massachusetts Amherst)
Abstract: Modern applications are end-to-end encrypted to prevent data from being read or secretly modified. 5G tech nology provides ubiquitous access to these applications without compromising the application-specific performance and latency goals. In this paper, we empirically demonstrate that 5G radio communication becomes the side channel to precisely infer the user's applications in real-time. The key idea lies in observing the 5G physical and MAC layer interactions over time that reveal the application's behavior. The MAC layer receives the data from the application and requests the network to assign the radio resource blocks. The network assigns the radio resources as per application requirements, such as priority, Quality of Service (QoS) needs, amount of data to be transmitted, and buffer size. The adversary can passively observe the radio resources to fingerprint the applications. We empirically demonstrate this attack by considering four different categories of applications: online shopping, voice/video conferencing, video streaming, and Over-The-Top (OTT) media platforms. Finally, we have also demonstrated that an attacker can differentiate various types of applications in real-time within each category.
Authors: Ziming Wang, Rebecka J\"ornsten
Abstract: Given a pair of point clouds, the goal of assembly is to recover a rigid transformation that aligns one point cloud to the other. This task is challenging because the point clouds may be non-overlapped, and they may have arbitrary initial positions. To address these difficulties, we propose a method, called SE(3)-bi-equivariant transformer (BITR), based on the SE(3)-bi-equivariance prior of the task: it guarantees that when the inputs are rigidly perturbed, the output will transform accordingly. Due to its equivariance property, BITR can not only handle non-overlapped PCs, but also guarantee robustness against initial positions. Specifically, BITR first extracts features of the inputs using a novel $SE(3) \times SE(3)$-transformer, and then projects the learned feature to group SE(3) as the output. Moreover, we theoretically show that swap and scale equivariances can be incorporated into BITR, thus it further guarantees stable performance under scaling and swapping the inputs. We experimentally show the effectiveness of BITR in practical tasks.
Authors: Ziheng Chen, Yue Song, Xiao-Jun Wu, Gaowen Liu, Nicu Sebe
Abstract: Global Covariance Pooling (GCP) has been demonstrated to improve the performance of Deep Neural Networks (DNNs) by exploiting second-order statistics of high-level representations. GCP typically performs classification of the covariance matrices by applying matrix function normalization, such as matrix logarithm or power, followed by a Euclidean classifier. However, covariance matrices inherently lie in a Riemannian manifold, known as the Symmetric Positive Definite (SPD) manifold. The current literature does not provide a satisfactory explanation of why Euclidean classifiers can be applied directly to Riemannian features after the normalization of the matrix power. To mitigate this gap, this paper provides a comprehensive and unified understanding of the matrix logarithm and power from a Riemannian geometry perspective. The underlying mechanism of matrix functions in GCP is interpreted from two perspectives: one based on tangent classifiers (Euclidean classifiers on the tangent space) and the other based on Riemannian classifiers. Via theoretical analysis and empirical validation through extensive experiments on fine-grained and large-scale visual classification datasets, we conclude that the working mechanism of the matrix functions should be attributed to the Riemannian classifiers they implicitly respect.
Authors: Hongyu Wang, Shuming Ma, Ruiping Wang, Furu Wei
Abstract: We introduce, Q-Sparse, a simple yet effective approach to training sparsely-activated large language models (LLMs). Q-Sparse enables full sparsity of activations in LLMs which can bring significant efficiency gains in inference. This is achieved by applying top-K sparsification to the activations and the straight-through-estimator to the training. We also introduce Block Q-Sparse for batch training and inference. The key results from this work are, (1) Q-Sparse can achieve results comparable to those of baseline LLMs while being much more efficient at inference time; (2) We present an inference-optimal scaling law for sparsely-activated LLMs; (3) Q-Sparse is effective in different settings, including training-from-scratch, continue-training of off-the-shelf LLMs, and finetuning; (4) Q-Sparse works for both full-precision and 1-bit LLMs (e.g., BitNet b1.58). Particularly, the synergy of BitNet b1.58 and Q-Sparse (can be equipped with MoE) provides the cornerstone and a clear path to revolutionize the efficiency, including cost and energy consumption, of future LLMs.
Authors: Christoph Minixhofer, Ond\v{r}ej Klejch, Peter Bell
Abstract: Many recently published Text-to-Speech (TTS) systems produce audio close to real speech. However, TTS evaluation needs to be revisited to make sense of the results obtained with the new architectures, approaches and datasets. We propose evaluating the quality of synthetic speech as a combination of multiple factors such as prosody, speaker identity, and intelligibility. Our approach assesses how well synthetic speech mirrors real speech by obtaining correlates of each factor and measuring their distance from both real speech datasets and noise datasets. We benchmark 35 TTS systems developed between 2008 and 2024 and show that our score computed as an unweighted average of factors strongly correlates with the human evaluations from each time period.
Authors: Audrey Huang, Wenhao Zhan, Tengyang Xie, Jason D. Lee, Wen Sun, Akshay Krishnamurthy, Dylan J. Foster
Abstract: Language model alignment methods, such as reinforcement learning from human feedback (RLHF), have led to impressive advances in language model capabilities, but existing techniques are limited by a widely observed phenomenon known as overoptimization, where the quality of the language model plateaus or degrades over the course of the alignment process. Overoptimization is often attributed to overfitting to an inaccurate reward model, and while it can be mitigated through online data collection, this is infeasible in many settings. This raises a fundamental question: Do existing offline alignment algorithms make the most of the data they have, or can their sample-efficiency be improved further? We address this question with a new algorithm for offline alignment, $\chi^2$-Preference Optimization ($\chi$PO). $\chi$PO is a one-line change to Direct Preference Optimization (DPO; Rafailov et al., 2023), which only involves modifying the logarithmic link function in the DPO objective. Despite this minimal change, $\chi$PO implicitly implements the principle of pessimism in the face of uncertainty via regularization with the $\chi^2$-divergence -- which quantifies uncertainty more effectively than KL-regularization -- and provably alleviates overoptimization, achieving sample-complexity guarantees based on single-policy concentrability -- the gold standard in offline reinforcement learning. $\chi$PO's simplicity and strong guarantees make it the first practical and general-purpose offline alignment algorithm that is provably robust to overoptimization.
Authors: Jing Xu, Yung-Cheng Hsu, William Biscarri
Abstract: Securities lending is an important part of the financial market structure, where agent lenders help long term institutional investors to lend out their securities to short sellers in exchange for a lending fee. Agent lenders within the market seek to optimize revenue by lending out securities at the highest rate possible. Typically, this rate is set by hard-coded business rules or standard supervised machine learning models. These approaches are often difficult to scale and are not adaptive to changing market conditions. Unlike a traditional stock exchange with a centralized limit order book, the securities lending market is organized similarly to an e-commerce marketplace, where agent lenders and borrowers can transact at any agreed price in a bilateral fashion. This similarity suggests that the use of typical methods for addressing dynamic pricing problems in e-commerce could be effective in the securities lending market. We show that existing contextual bandit frameworks can be successfully utilized in the securities lending market. Using offline evaluation on real historical data, we show that the contextual bandit approach can consistently outperform typical approaches by at least 15% in terms of total revenue generated.
Authors: Wenxuan Lu, Mengshun Hu, Yansheng Qiu, Liang Liao, Zheng Wang
Abstract: With the development of VR-related techniques, viewers can enjoy a realistic and immersive experience through a head-mounted display, while omnidirectional video with a low frame rate can lead to user dizziness. However, the prevailing plane frame interpolation methodologies are unsuitable for Omnidirectional Video Interpolation, chiefly due to the lack of models tailored to such videos with strong distortion, compounded by the scarcity of valuable datasets for Omnidirectional Video Frame Interpolation. In this paper, we introduce the benchmark dataset, 360VFI, for Omnidirectional Video Frame Interpolation. We present a practical implementation that introduces a distortion prior from omnidirectional video into the network to modulate distortions. We especially propose a pyramid distortion-sensitive feature extractor that uses the unique characteristics of equirectangular projection (ERP) format as prior information. Moreover, we devise a decoder that uses an affine transformation to facilitate the synthesis of intermediate frames further. 360VFI is the first dataset and benchmark that explores the challenge of Omnidirectional Video Frame Interpolation. Through our benchmark analysis, we presented four different distortion conditions scenes in the proposed 360VFI dataset to evaluate the challenge triggered by distortion during interpolation. Besides, experimental results demonstrate that Omnidirectional Video Interpolation can be effectively improved by modeling for omnidirectional distortion.