Authors: Stamatis Mastromichalakis
Activation functions (AFs) are crucial components of deep neural networks (DNNs), having a significant impact on their performance. An activation function in a DNN is typically a smooth, nonlinear function that transforms an input signal into an output signal for the subsequent layer. In this paper, we propose the Parametric Leaky Tanh (PLTanh), a novel hybrid activation function designed to combine the strengths of both the Tanh and Leaky ReLU (LReLU) activation functions. PLTanh is differentiable at all points and addresses the 'dying ReLU' problem by ensuring a non-zero gradient for negative inputs, consistent with the behavior of LReLU. By integrating the unique advantages of these two diverse activation functions, PLTanh facilitates the learning of more intricate nonlinear relationships within the network. This paper presents an empirical evaluation of PLTanh against established activation functions, namely ReLU, LReLU, and ALReLU utilizing five diverse datasets.
Authors: Hsuan-Kung Yang, Tsung-Chih Chiang, Ting-Ru Liu, Chun-Wei Huang, Jou-Min Liu, Chun-Yi Lee
The challenge of navigation in environments with dynamic objects continues to be a central issue in the study of autonomous agents. While predictive methods hold promise, their reliance on precise state information makes them less practical for real-world implementation. This study presents visual forecasting as an innovative alternative. By introducing intuitive visual cues, this approach projects the future trajectories of dynamic objects to improve agent perception and enable anticipatory actions. Our research explores two distinct strategies for conveying predictive information through visual forecasting: (1) sequences of bounding boxes, and (2) augmented paths. To validate the proposed visual forecasting strategies, we initiate evaluations in simulated environments using the Unity engine and then extend these evaluations to real-world scenarios to assess both practicality and effectiveness. The results confirm the viability of visual forecasting as a promising solution for navigation and obstacle avoidance in dynamic environments.
Authors: Girik Malik, Dakarai Crowder, Ennio Mingolla
Adversarial attacks can affect the object recognition capabilities of machines in wild. These can often result from spurious correlations between input and class labels, and are prone to memorization in large networks. While networks are expected to do automated feature selection, it is not effective at the scale of the object. Humans, however, are able to select the minimum set of features required to form a robust representation of an object. In this work, we show that finetuning any pretrained off-the-shelf network with Extreme Image Transformations (EIT) not only helps in learning a robust latent representation, it also improves the performance of these networks against common adversarial attacks of various intensities. Our EIT trained networks show strong activations in the object regions even when tested with more intense noise, showing promising generalizations across different kinds of adversarial attacks.
Authors: Tianji Cong, Madelon Hulsebos, Zhenjie Sun, Paul Groth, H. V. Jagadish
Language models and specialized table embedding models have recently demonstrated strong performance on many tasks over tabular data. Researchers and practitioners are keen to leverage these models in many new application contexts; but limited understanding of the strengths and weaknesses of these models, and the table representations they generate, makes the process of finding a suitable model for a given task reliant on trial and error. There is an urgent need to gain a comprehensive understanding of these models to minimize inefficiency and failures in downstream usage.
To address this need, we propose Observatory, a formal framework to systematically analyze embedding representations of relational tables. Motivated both by invariants of the relational data model and by statistical considerations regarding data distributions, we define eight primitive properties, and corresponding measures to quantitatively characterize table embeddings for these properties. Based on these properties, we define an extensible framework to evaluate language and table embedding models. We collect and synthesize a suite of datasets and use Observatory to analyze seven such models. Our analysis provides insights into the strengths and weaknesses of learned representations over tables. We find, for example, that some models are sensitive to table structure such as column order, that functional dependencies are rarely reflected in embeddings, and that specialized table embedding models have relatively lower sample fidelity. Such insights help researchers and practitioners better anticipate model behaviors and select appropriate models for their downstream tasks, while guiding researchers in the development of new models.
Authors: Matthew J. Baumstark, Giuseppe Vinci
The classification of galaxy morphologies is an important step in the investigation of theories of hierarchical structure formation. While human expert visual classification remains quite effective and accurate, it cannot keep up with the massive influx of data from emerging sky surveys. A variety of approaches have been proposed to classify large numbers of galaxies; these approaches include crowdsourced visual classification, and automated and computational methods, such as machine learning methods based on designed morphology statistics and deep learning. In this work, we develop two novel galaxy morphology statistics, descent average and descent variance, which can be efficiently extracted from telescope galaxy images. We further propose simplified versions of the existing image statistics concentration, asymmetry, and clumpiness, which have been widely used in the literature of galaxy morphologies. We utilize the galaxy image data from the Sloan Digital Sky Survey to demonstrate the effective performance of our proposed image statistics at accurately detecting spiral and elliptical galaxies when used as features of a random forest classifier.
Authors: Gregory Palmer, Chris Parry, Daniel J.B. Harrold, Chris Willis
The rapid increase in the number of cyber-attacks in recent years raises the need for principled methods for defending networks against malicious actors. Deep reinforcement learning (DRL) has emerged as a promising approach for mitigating these attacks. However, while DRL has shown much potential for cyber-defence, numerous challenges must be overcome before DRL can be applied to autonomous cyber-operations (ACO) at scale. Principled methods are required for environments that confront learners with very high-dimensional state spaces, large multi-discrete action spaces, and adversarial learning. Recent works have reported success in solving these problems individually. There have also been impressive engineering efforts towards solving all three for real-time strategy games. However, applying DRL to the full ACO problem remains an open challenge. Here, we survey the relevant DRL literature and conceptualize an idealised ACO-DRL agent. We provide: i.) A summary of the domain properties that define the ACO problem; ii.) A comprehensive evaluation of the extent to which domains used for benchmarking DRL approaches are comparable to ACO; iii.) An overview of state-of-the-art approaches for scaling DRL to domains that confront learners with the curse of dimensionality, and; iv.) A survey and critique of current methods for limiting the exploitability of agents within adversarial settings from the perspective of ACO. We conclude with open research questions that we hope will motivate future directions for researchers and practitioners working on ACO.
Authors: Hao Sun, Alihan Hüyük, Daniel Jarrett, Mihaela van der Schaar
Learning transparent, interpretable controllers with offline data in decision-making systems is an essential area of research due to its potential to reduce the risk of applications in real-world systems. However, in responsibility-sensitive settings such as healthcare, decision accountability is of paramount importance, yet has not been adequately addressed by the literature. This paper introduces the Accountable Offline Controller (AOC) that employs the offline dataset as the Decision Corpus and performs accountable control based on a tailored selection of examples, referred to as the Corpus Subset. ABC operates effectively in low-data scenarios, can be extended to the strictly offline imitation setting, and displays qualities of both conservation and adaptability. We assess ABC's performance in both simulated and real-world healthcare scenarios, emphasizing its capability to manage offline control tasks with high levels of performance while maintaining accountability.
Keywords: Interpretable Reinforcement Learning, Explainable Reinforcement Learning, Reinforcement Learning Transparency, Offline Reinforcement Learning, Batched Control.
Authors: Yi Sui, Tongzi Wu, Jesse C. Cresswell, Ga Wu, George Stein, Xiao Shi Huang, Xiaochen Zhang, Maksims Volkovs
Self-supervised representation learning~(SSRL) has advanced considerably by exploiting the transformation invariance assumption under artificially designed data augmentations. While augmentation-based SSRL algorithms push the boundaries of performance in computer vision and natural language processing, they are often not directly applicable to other data modalities, and can conflict with application-specific data augmentation constraints. This paper presents an SSRL approach that can be applied to any data modality and network architecture because it does not rely on augmentations or masking. Specifically, we show that high-quality data representations can be learned by reconstructing random data projections. We evaluate the proposed approach on a wide range of representation learning tasks that span diverse modalities and real-world applications. We show that it outperforms multiple state-of-the-art SSRL baselines. Due to its wide applicability and strong empirical results, we argue that learning from randomness is a fruitful research direction worthy of attention and further study.
Authors: Hannah Day, Yonatan Kahn, Daniel A. Roberts
Fully-connected deep neural networks with weights initialized from independent Gaussian distributions can be tuned to criticality, which prevents the exponential growth or decay of signals propagating through the network. However, such networks still exhibit fluctuations that grow linearly with the depth of the network, which may impair the training of networks with width comparable to depth. We show analytically that rectangular networks with tanh activations and weights initialized from the ensemble of orthogonal matrices have corresponding preactivation fluctuations which are independent of depth, to leading order in inverse width. Moreover, we demonstrate numerically that, at initialization, all correlators involving the neural tangent kernel (NTK) and its descendants at leading order in inverse width -- which govern the evolution of observables during training -- saturate at a depth of $\sim 20$, rather than growing without bound as in the case of Gaussian initializations. We speculate that this structure preserves finite-width feature learning while reducing overall noise, thus improving both generalization and training speed. We provide some experimental justification by relating empirical measurements of the NTK to the superior performance of deep nonlinear orthogonal networks trained under full-batch gradient descent on the MNIST and CIFAR-10 classification tasks.
Authors: Linbo Liu, Trong Nghia Hoang, Lam M. Nguyen, Tsui-Wei Weng
Randomized smoothing has recently attracted attentions in the field of adversarial robustness to provide provable robustness guarantees on smoothed neural network classifiers. However, existing works show that vanilla randomized smoothing usually does not provide good robustness performance and often requires (re)training techniques on the base classifier in order to boost the robustness of the resulting smoothed classifier. In this work, we propose two cost-effective approaches to boost the robustness of randomized smoothing while preserving its clean performance. The first approach introduces a new robust training method AdvMacerwhich combines adversarial training and robustness certification maximization for randomized smoothing. We show that AdvMacer can improve the robustness performance of randomized smoothing classifiers compared to SOTA baselines, while being 3x faster to train than MACER baseline. The second approach introduces a post-processing method EsbRS which greatly improves the robustness certificate based on building model ensembles. We explore different aspects of model ensembles that has not been studied by prior works and propose a novel design methodology to further improve robustness of the ensemble based on our theoretical analysis.
Authors: Zheqing Zhu, Yueyang Liu, Xu Kuang, Benjamin Van Roy
Real-world applications of contextual bandits often exhibit non-stationarity due to seasonality, serendipity, and evolving social trends. While a number of non-stationary contextual bandit learning algorithms have been proposed in the literature, they excessively explore due to a lack of prioritization for information of enduring value, or are designed in ways that do not scale in modern applications with high-dimensional user-specific features and large action set, or both. In this paper, we introduce a novel non-stationary contextual bandit algorithm that addresses these concerns. It combines a scalable, deep-neural-network-based architecture with a carefully designed exploration mechanism that strategically prioritizes collecting information with the most lasting value in a non-stationary environment. Through empirical evaluations on two real-world recommendation datasets, which exhibit pronounced non-stationarity, we demonstrate that our approach significantly outperforms the state-of-the-art baselines.
Authors: Philip Wong, Phue Thant, Pratiksha Yadav, Ruta Antaliya, Jongwook Woo
This paper discusses predictive performance and processes undertaken on flight pricing data utilizing r2(r-square) and RMSE that leverages a large dataset, originally from Expedia.com, consisting of approximately 20 million records or 4.68 gigabytes. The project aims to determine the best models usable in the real world to predict airline ticket fares for non-stop flights across the US. Therefore, good generalization capability and optimized processing times are important measures for the model.
We will discover key business insights utilizing feature importance and discuss the process and tools used for our analysis. Four regression machine learning algorithms were utilized: Random Forest, Gradient Boost Tree, Decision Tree, and Factorization Machines utilizing Cross Validator and Training Validator functions for assessing performance and generalization capability.
Authors: Ruotong Liao, Xu Jia, Yunpu Ma, Volker Tresp
The rapid advancements in large language models (LLMs) have ignited interest in the temporal knowledge graph (tKG) domain, where conventional carefully designed embedding-based and rule-based models dominate. The question remains open of whether pre-trained LLMs can understand structured temporal relational data and replace them as the foundation model for temporal relational forecasting. Therefore, we bring temporal knowledge forecasting into the generative setting. However, challenges occur in the huge chasms between complex temporal graph data structure and sequential natural expressions LLMs can handle, and between the enormous data sizes of tKGs and heavy computation costs of finetuning LLMs. To address these challenges, we propose a novel retrieval augmented generation framework that performs generative forecasting on tKGs named GenTKG, which combines a temporal logical rule-based retrieval strategy and lightweight parameter-efficient instruction tuning. Extensive experiments have shown that GenTKG outperforms conventional methods of temporal relational forecasting under low computation resources. GenTKG also highlights remarkable transferability with exceeding performance on unseen datasets without re-training. Our work reveals the huge potential of LLMs in the tKG domain and opens a new frontier for generative forecasting on tKGs.
Authors: Changhe Chen, Mozhgan Pourkeshavarz, Amir Rasouli
Benchmarking is a common method for evaluating trajectory prediction models for autonomous driving. Existing benchmarks rely on datasets, which are biased towards more common scenarios, such as cruising, and distance-based metrics that are computed by averaging over all scenarios. Following such a regiment provides a little insight into the properties of the models both in terms of how well they can handle different scenarios and how admissible and diverse their outputs are. There exist a number of complementary metrics designed to measure the admissibility and diversity of trajectories, however, they suffer from biases, such as length of trajectories.
In this paper, we propose a new benChmarking paRadIgm for evaluaTing trajEctoRy predIction Approaches (CRITERIA). Particularly, we propose 1) a method for extracting driving scenarios at varying levels of specificity according to the structure of the roads, models' performance, and data properties for fine-grained ranking of prediction models; 2) A set of new bias-free metrics for measuring diversity, by incorporating the characteristics of a given scenario, and admissibility, by considering the structure of roads and kinematic compliancy, motivated by real-world driving constraints. 3) Using the proposed benchmark, we conduct extensive experimentation on a representative set of the prediction models using the large scale Argoverse dataset. We show that the proposed benchmark can produce a more accurate ranking of the models and serve as a means of characterizing their behavior. We further present ablation studies to highlight contributions of different elements that are used to compute the proposed metrics.
Authors: Zhongji Zhang, Yuhang Wang, Yinghao Zhu, Xinyu Ma, Tianlong Wang, Chaohe Zhang, Yasha Wang, Liantao Ma
Due to the limited information about emerging diseases, symptoms are hard to be noticed and recognized, so that the window for clinical intervention could be ignored. An effective prognostic model is expected to assist doctors in making right diagnosis and designing personalized treatment plan, so to promptly prevent unfavorable outcomes. However, in the early stage of a disease, limited data collection and clinical experiences, plus the concern out of privacy and ethics, may result in restricted data availability for reference, to the extent that even data labels are difficult to mark correctly. In addition, Electronic Medical Record (EMR) data of different diseases or of different sources of the same disease can prove to be having serious cross-dataset feature misalignment problems, greatly mutilating the efficiency of deep learning models. This article introduces a transfer learning method to build a transition model from source dataset to target dataset. By way of constraining the distribution shift of features generated in disparate domains, domain-invariant features that are exclusively relative to downstream tasks are captured, so to cultivate a unified domain-invariant encoder across various task domains to achieve better feature representation. Experimental results of several target tasks demonstrate that our proposed model outperforms competing baseline methods and has higher rate of training convergence, especially in dealing with limited data amount. A multitude of experiences have proven the efficacy of our method to provide more accurate predictions concerning newly emergent pandemics and other diseases.
Authors: Bahareh Nikpour, Narges Armanfard
Attention mechanisms have exhibited promising potential in enhancing learning models by identifying salient portions of input data. This is particularly valuable in scenarios where limited training samples are accessible due to challenges in data collection and labeling. Drawing inspiration from human recognition processes, we posit that an AI baseline's performance could be more accurate and dependable if it is exposed to essential segments of raw data rather than the entire input dataset, akin to human perception. However, the task of selecting these informative data segments, referred to as hard attention finding, presents a formidable challenge. In situations with few training samples, existing studies struggle to locate such informative regions due to the large number of training parameters that cannot be effectively learned from the available limited samples. In this study, we introduce a novel and practical framework for achieving explainable hard attention finding, specifically tailored for few-shot learning scenarios, called FewXAT. Our approach employs deep reinforcement learning to implement the concept of hard attention, directly impacting raw input data and thus rendering the process interpretable for human understanding. Through extensive experimentation across various benchmark datasets, we demonstrate the efficacy of our proposed method.
Authors: Tianrong Chen, Jiatao Gu, Laurent Dinh, Evangelos A. Theodorou, Josh Susskind, Shuangfei Zhai
Diffusion models (DMs) represent state-of-the-art generative models for continuous inputs. DMs work by constructing a Stochastic Differential Equation (SDE) in the input space (ie, position space), and using a neural network to reverse it. In this work, we introduce a novel generative modeling framework grounded in \textbf{phase space dynamics}, where a phase space is defined as {an augmented space encompassing both position and velocity.} Leveraging insights from Stochastic Optimal Control, we construct a path measure in the phase space that enables efficient sampling. {In contrast to DMs, our framework demonstrates the capability to generate realistic data points at an early stage of dynamics propagation.} This early prediction sets the stage for efficient data generation by leveraging additional velocity information along the trajectory. On standard image generation benchmarks, our model yields favorable performance over baselines in the regime of small Number of Function Evaluations (NFEs). Furthermore, our approach rivals the performance of diffusion models equipped with efficient sampling techniques, underscoring its potential as a new tool generative modeling.
Authors: Ensiye Kiyamousavi, Boris Kraychev, Ivan Koychev
Federated learning (FL) is a decentralized machine learning approach where independent learners process data privately. Its goal is to create a robust and accurate model by aggregating and retraining local models over multiple rounds. However, FL faces challenges regarding data heterogeneity and model aggregation effectiveness. In order to simulate real-world data, researchers use methods for data partitioning that transform a dataset designated for centralized learning into a group of sub-datasets suitable for distributed machine learning with different data heterogeneity. In this paper, we study the currently popular data partitioning techniques and visualize their main disadvantages: the lack of precision in the data diversity, which leads to unreliable heterogeneity indexes, and the inability to incrementally challenge the FL algorithms. To resolve this problem, we propose a method that leverages entropy and symmetry to construct 'the most challenging' and controllable data distributions with gradual difficulty. We introduce a metric to measure data heterogeneity among the learning agents and a transformation technique that divides any dataset into splits with precise data diversity. Through a comparative study, we demonstrate the superiority of our method over existing FL data partitioning approaches, showcasing its potential to challenge model aggregation algorithms. Experimental results indicate that our approach gradually challenges the FL strategies, and the models trained on FedSym distributions are more distinct.
Authors: Gellért Weisz, András György, Csaba Szepesvári
We consider online reinforcement learning (RL) in episodic Markov decision processes (MDPs) under the linear $q^\pi$-realizability assumption, where it is assumed that the action-values of all policies can be expressed as linear functions of state-action features. This class is known to be more general than linear MDPs, where the transition kernel and the reward function are assumed to be linear functions of the feature vectors. As our first contribution, we show that the difference between the two classes is the presence of states in linearly $q^\pi$-realizable MDPs where for any policy, all the actions have approximately equal values, and skipping over these states by following an arbitrarily fixed policy in those states transforms the problem to a linear MDP. Based on this observation, we derive a novel (computationally inefficient) learning algorithm for linearly $q^\pi$-realizable MDPs that simultaneously learns what states should be skipped over and runs another learning algorithm on the linear MDP hidden in the problem. The method returns an $\epsilon$-optimal policy after $\text{polylog}(H, d)/\epsilon^2$ interactions with the MDP, where $H$ is the time horizon and $d$ is the dimension of the feature vectors, giving the first polynomial-sample-complexity online RL algorithm for this setting. The results are proved for the misspecified case, where the sample complexity is shown to degrade gracefully with the misspecification error.
Authors: Arman Maesumi, Paul Guerrero, Vladimir G. Kim, Matthew Fisher, Siddhartha Chaudhuri, Noam Aigerman, Daniel Ritchie
Exploring variations of 3D shapes is a time-consuming process in traditional 3D modeling tools. Deep generative models of 3D shapes often feature continuous latent spaces that can, in principle, be used to explore potential variations starting from a set of input shapes. In practice, doing so can be problematic: latent spaces are high dimensional and hard to visualize, contain shapes that are not relevant to the input shapes, and linear paths through them often lead to sub-optimal shape transitions. Furthermore, one would ideally be able to explore variations in the original high-quality meshes used to train the generative model, not its lower-quality output geometry. In this paper, we present a method to explore variations among a given set of landmark shapes by constructing a mapping from an easily-navigable 2D exploration space to a subspace of a pre-trained generative model. We first describe how to find a mapping that spans the set of input landmark shapes and exhibits smooth variations between them. We then show how to turn the variations in this subspace into deformation fields, to transfer those variations to high-quality meshes for the landmark shapes. Our results show that our method can produce visually-pleasing and easily-navigable 2D exploration spaces for several different shape categories, especially as compared to prior work on learning deformation spaces for 3D shapes.
Authors: Bowen Jin, Hansi Zeng, Guoyin Wang, Xiusi Chen, Tianxin Wei, Ruirui Li, Zhengyang Wang, Zheng Li, Yang Li, Hanqing Lu, Suhang Wang, Jiawei Han, Xianfeng Tang
Semantic identifier (ID) is an important concept in information retrieval that aims to preserve the semantics of objects such as documents and items inside their IDs. Previous studies typically adopt a two-stage pipeline to learn semantic IDs by first procuring embeddings using off-the-shelf text encoders and then deriving IDs based on the embeddings. However, each step introduces potential information loss and there is usually an inherent mismatch between the distribution of embeddings within the latent space produced by text encoders and the anticipated distribution required for semantic indexing. Nevertheless, it is non-trivial to design a method that can learn the document's semantic representations and its hierarchical structure simultaneously, given that semantic IDs are discrete and sequentially structured, and the semantic supervision is deficient. In this paper, we introduce LMINDEXER, a self-supervised framework to learn semantic IDs with a generative language model. We tackle the challenge of sequential discrete ID by introducing a semantic indexer capable of generating neural sequential discrete representations with progressive training and contrastive learning. In response to the semantic supervision deficiency, we propose to train the model with a self-supervised document reconstruction objective. The learned semantic indexer can facilitate various downstream tasks, such as recommendation and retrieval. We conduct experiments on three tasks including recommendation, product search, and document retrieval on five datasets from various domains, where LMINDEXER outperforms competitive baselines significantly and consistently.
Authors: Andreas Madsen, Siva Reddy, Sarath Chandar
A common approach to explain NLP models, is to use importance measures that express which tokens are important for a prediction. Unfortunately, such explanations are often wrong despite being persuasive. Therefore, it is essential to measure their faithfulness. One such metric is if tokens are truly important, then masking them should result in worse model performance. However, token masking introduces out-of-distribution issues and existing solutions are computationally expensive and employ proxy-models. Furthermore, other metrics are very limited in scope. In this work, we propose an inherently faithfulness measurable model that addresses these challenges. This is achieved by using a novel fine-tuning method that incorporates masking, such that masking tokens become in-distribution by design. This differs from existing approaches, which are completely model-agnostic but are inapplicable in practice. We demonstrate the generality of our approach by applying it to various tasks and validate it using statistical in-distribution tests. Additionally, because masking is in-distribution, importance measures which themselves use masking become more faithful, thus our model becomes more explainable.
Authors: Nate Gruver, Marc Finzi, Shikai Qiu, Andrew Gordon Wilson
By encoding time series as a string of numerical digits, we can frame time series forecasting as next-token prediction in text. Developing this approach, we find that large language models (LLMs) such as GPT-3 and LLaMA-2 can surprisingly zero-shot extrapolate time series at a level comparable to or exceeding the performance of purpose-built time series models trained on the downstream tasks. To facilitate this performance, we propose procedures for effectively tokenizing time series data and converting discrete distributions over tokens into highly flexible densities over continuous values. We argue the success of LLMs for time series stems from their ability to naturally represent multimodal distributions, in conjunction with biases for simplicity, and repetition, which align with the salient features in many time series, such as repeated seasonal trends. We also show how LLMs can naturally handle missing data without imputation through non-numerical text, accommodate textual side information, and answer questions to help explain predictions. While we find that increasing model size generally improves performance on time series, we show GPT-4 can perform worse than GPT-3 because of how it tokenizes numbers, and poor uncertainty calibration, which is likely the result of alignment interventions such as RLHF.
Authors: Sia Gholami, Marwan Omar
Natural Language Processing (NLP) has undergone transformative changes with the advent of deep learning methodologies. One challenge persistently confronting researchers is the scarcity of high-quality, annotated datasets that drive these models. This paper explores the nuances of synthetic data generation in NLP, with a focal point on template-based question generation. By assessing its advantages, including data augmentation potential and the introduction of structured variety, we juxtapose these benefits against inherent limitations, such as the risk of overfitting and the constraints posed by pre-defined templates. Drawing from empirical evaluations, we demonstrate the impact of template-based synthetic data on the performance of modern transformer models. We conclude by emphasizing the delicate balance required between synthetic and real-world data, and the future trajectories of integrating synthetic data in model training pipelines. The findings aim to guide NLP practitioners in harnessing synthetic data's potential, ensuring optimal model performance in diverse applications.
Authors: Aaron Defazio, Ashok Cutkosky, Harsh Mehta, Konstantin Mishchenko
Learning rate schedules used in practice bear little resemblance to those recommended by theory. We close much of this theory/practice gap, and as a consequence are able to derive new problem-adaptive learning rate schedules. Our key technical contribution is a refined analysis of learning rate schedules for a wide class of optimization algorithms (including SGD). In contrast to most prior works that study the convergence of the average iterate, we study the last iterate, which is what most people use in practice. When considering only worst-case analysis, our theory predicts that the best choice is the linear decay schedule: a popular choice in practice that sets the stepsize proportionally to $1 - t/T$, where $t$ is the current iteration and $T$ is the total number of steps. To go beyond this worst-case analysis, we use the observed gradient norms to derive schedules refined for any particular task. These refined schedules exhibit learning rate warm-up and rapid learning rate annealing near the end of training. Ours is the first systematic approach to automatically yield both of these properties. We perform the most comprehensive evaluation of learning rate schedules to date, evaluating across 10 diverse deep learning problems, a series of LLMs, and a suite of logistic regression problems. We validate that overall, the linear-decay schedule matches or outperforms all commonly used default schedules including cosine annealing, and that our schedule refinement method gives further improvements.
Authors: Mingyang Deng, Lucas Tao, Joe Benton
Recent works have proposed that activations in language models can be modelled as sparse linear combinations of vectors corresponding to features of input text. Under this assumption, these works aimed to reconstruct feature directions using sparse coding. We develop metrics to assess the success of these sparse coding techniques and test the validity of the linearity and sparsity assumptions. We show our metrics can predict the level of sparsity on synthetic sparse linear activations, and can distinguish between sparse linear data and several other distributions. We use our metrics to measure levels of sparsity in several language models. We find evidence that language model activations can be accurately modelled by sparse linear combinations of features, significantly more so than control datasets. We also show that model activations appear to be sparsest in the first and final layers.
Authors: Qingyue Zhao, Banghua Zhu
We characterize the statistical efficiency of knowledge transfer through $n$ samples from a teacher to a probabilistic student classifier with input space $\mathcal S$ over labels $\mathcal A$. We show that privileged information at three progressive levels accelerates the transfer. At the first level, only samples with hard labels are known, via which the maximum likelihood estimator attains the minimax rate $\sqrt{{|{\mathcal S}||{\mathcal A}|}/{n}}$. The second level has the teacher probabilities of sampled labels available in addition, which turns out to boost the convergence rate lower bound to ${{|{\mathcal S}||{\mathcal A}|}/{n}}$. However, under this second data acquisition protocol, minimizing a naive adaptation of the cross-entropy loss results in an asymptotically biased student. We overcome this limitation and achieve the fundamental limit by using a novel empirical variant of the squared error logit loss. The third level further equips the student with the soft labels (complete logits) on ${\mathcal A}$ given every sampled input, thereby provably enables the student to enjoy a rate ${|{\mathcal S}|}/{n}$ free of $|{\mathcal A}|$. We find any Kullback-Leibler divergence minimizer to be optimal in the last case. Numerical simulations distinguish the four learners and corroborate our theory.
Authors: Saptarshi Roy, Ambuj Tewari
We consider the problem of model selection in a high-dimensional sparse linear regression model under the differential privacy framework. In particular, we consider the problem of differentially private best subset selection and study its utility guarantee. We adopt the well-known exponential mechanism for selecting the best model, and under a certain margin condition, we establish its strong model recovery property. However, the exponential search space of the exponential mechanism poses a serious computational bottleneck. To overcome this challenge, we propose a Metropolis-Hastings algorithm for the sampling step and establish its polynomial mixing time to its stationary distribution in the problem parameters $n,p$, and $s$. Furthermore, we also establish approximate differential privacy for the final estimates of the Metropolis-Hastings random walk using its mixing property. Finally, we also perform some illustrative simulations that echo the theoretical findings of our main results.
Authors: Tim Lebailly, Thomas Stegmüller, Behzad Bozorgtabar, Jean-Philippe Thiran, Tinne Tuytelaars
Leveraging nearest neighbor retrieval for self-supervised representation learning has proven beneficial with object-centric images. However, this approach faces limitations when applied to scene-centric datasets, where multiple objects within an image are only implicitly captured in the global representation. Such global bootstrapping can lead to undesirable entanglement of object representations. Furthermore, even object-centric datasets stand to benefit from a finer-grained bootstrapping approach. In response to these challenges, we introduce a novel Cross-Image Object-Level Bootstrapping method tailored to enhance dense visual representation learning. By employing object-level nearest neighbor bootstrapping throughout the training, CrIBo emerges as a notably strong and adequate candidate for in-context learning, leveraging nearest neighbor retrieval at test time. CrIBo shows state-of-the-art performance on the latter task while being highly competitive in more standard downstream segmentation tasks. Our code and pretrained models will be publicly available upon acceptance.
Authors: Ankit Kulshrestha, Danylo Lykov, Ilya Safro, Yuri Alexeev
The current era of quantum computing has yielded several algorithms that promise high computational efficiency. While the algorithms are sound in theory and can provide potentially exponential speedup, there is little guidance on how to design proper quantum circuits to realize the appropriate unitary transformation to be applied to the input quantum state. In this paper, we present \texttt{QArchSearch}, an AI based quantum architecture search package with the \texttt{QTensor} library as a backend that provides a principled and automated approach to finding the best model given a task and input quantum state. We show that the search package is able to efficiently scale the search to large quantum circuits and enables the exploration of more complex models for different quantum applications. \texttt{QArchSearch} runs at scale and high efficiency on high-performance computing systems using a two-level parallelization scheme on both CPUs and GPUs, which has been demonstrated on the Polaris supercomputer.
Authors: Christos Boutsikas, Petros Drineas, Marios Mertzanidis, Alexandros Psomas, Paritosh Verma
We consider the problem of a revenue-maximizing seller with a large number of items $m$ for sale to $n$ strategic bidders, whose valuations are drawn independently from high-dimensional, unknown prior distributions. It is well-known that optimal and even approximately-optimal mechanisms for this setting are notoriously difficult to characterize or compute, and, even when they can be found, are often rife with various counter-intuitive properties. In this paper, following a model introduced recently by Cai and Daskalakis~\cite{cai2022recommender}, we consider the case that bidders' prior distributions can be well-approximated by a topic model. We design an active learning component, responsible for interacting with the bidders and outputting low-dimensional approximations of their types, and a mechanism design component, responsible for robustifying mechanisms for the low-dimensional model to work for the approximate types of the former component. On the active learning front, we cast our problem in the framework of Randomized Linear Algebra (RLA) for regression problems, allowing us to import several breakthrough results from that line of research, and adapt them to our setting. On the mechanism design front, we remove many restrictive assumptions of prior work on the type of access needed to the underlying distributions and the associated mechanisms. To the best of our knowledge, our work is the first to formulate connections between mechanism design, and RLA for active learning of regression problems, opening the door for further applications of randomized linear algebra primitives to mechanism design.
Authors: Gus Eggert, Kevin Huo, Mike Biven, Justin Waugh
It is well-established that large, diverse datasets play a pivotal role in the performance of modern AI systems for text and image modalities. However, there are no datasets for tabular data of comparable size and diversity to those available for text and images. Thus we present "TabLib'', a compilation of 627 million tables totaling 69 TiB, along with 867B tokens of context. TabLib was extracted from numerous file formats, including CSV, HTML, SQLite, PDF, Excel, and others, sourced from GitHub and Common Crawl. The size and diversity of TabLib offer considerable promise in the table modality, reminiscent of the original promise of foundational datasets for text and images, such as The Pile and LAION.
Authors: Nawras Alkassab, Chin-Tser Huang, Tania Lorido Botran
Content Delivery Networks carry the majority of Internet traffic, and the increasing demand for video content as a major IP traffic across the Internet highlights the importance of caching and prefetching optimization algorithms. Prefetching aims to make data available in the cache before the requester places its request to reduce access time and improve the Quality of Experience on the user side. Prefetching is well investigated in operating systems, compiler instructions, in-memory cache, local storage systems, high-speed networks, and cloud systems. Traditional prefetching techniques are well adapted to a particular access pattern, but fail to adapt to sudden variations or randomization in workloads. This paper explores the use of reinforcement learning to tackle the changes in user access patterns and automatically adapt over time. To this end, we propose, DeePref, a Deep Reinforcement Learning agent for online video content prefetching in Content Delivery Networks. DeePref is a prefetcher implemented on edge networks and is agnostic to hardware design, operating systems, and applications. Our results show that DeePref DRQN, using a real-world dataset, achieves a 17% increase in prefetching accuracy and a 28% increase in prefetching coverage on average compared to baseline approaches that use video content popularity as a building block to statically or dynamically make prefetching decisions. We also study the possibility of transfer learning of statistical models from one edge network into another, where unseen user requests from unknown distribution are observed. In terms of transfer learning, the increase in prefetching accuracy and prefetching coverage are [$30%$, $10%$], respectively. Our source code will be available on Github.
Authors: Thomas Decker, Ralf Gross, Alexander Koebler, Michael Lebacher, Ronald Schnitzer, Stefan H. Weber
In this paper, we investigate the practical relevance of explainable artificial intelligence (XAI) with a special focus on the producing industries and relate them to the current state of academic XAI research. Our findings are based on an extensive series of interviews regarding the role and applicability of XAI along the Machine Learning (ML) lifecycle in current industrial practice and its expected relevance in the future. The interviews were conducted among a great variety of roles and key stakeholders from different industry sectors. On top of that, we outline the state of XAI research by providing a concise review of the relevant literature. This enables us to provide an encompassing overview covering the opinions of the surveyed persons as well as the current state of academic research. By comparing our interview results with the current research approaches we reveal several discrepancies. While a multitude of different XAI approaches exists, most of them are centered around the model evaluation phase and data scientists. Their versatile capabilities for other stages are currently either not sufficiently explored or not popular among practitioners. In line with existing work, our findings also confirm that more efforts are needed to enable also non-expert users' interpretation and understanding of opaque AI models with existing methods and frameworks.
Authors: Chenzhong Yin, Mingxi Cheng, Xiongye Xiao, Xinghe Chen, Shahin Nazarian, Andrei Irimia, Paul Bogdan
The collective behavior of a network with heterogeneous, resource-limited information processing units (e.g., group of fish, flock of birds, or network of neurons) demonstrates high self-organization and complexity. These emergent properties arise from simple interaction rules where certain individuals can exhibit leadership-like behavior and influence the collective activity of the group. Motivated by the intricacy of these collectives, we propose a neural network (NN) architecture inspired by the rules observed in nature's collective ensembles. This NN structure contains workers that encompass one or more information processing units (e.g., neurons, filters, layers, or blocks of layers). Workers are either leaders or followers, and we train a leader-follower neural network (LFNN) by leveraging local error signals and optionally incorporating backpropagation (BP) and global loss. We investigate worker behavior and evaluate LFNNs through extensive experimentation. Our LFNNs trained with local error signals achieve significantly lower error rates than previous BP-free algorithms on MNIST and CIFAR-10 and even surpass BP-enabled baselines. In the case of ImageNet, our LFNN-l demonstrates superior scalability and outperforms previous BP-free algorithms by a significant margin.
Authors: Behrad Moniri, Donghwan Lee, Hamed Hassani, Edgar Dobriban
Feature learning is thought to be one of the fundamental reasons for the success of deep neural networks. It is rigorously known that in two-layer fully-connected neural networks under certain conditions, one step of gradient descent on the first layer followed by ridge regression on the second layer can lead to feature learning; characterized by the appearance of a separated rank-one component -- spike -- in the spectrum of the feature matrix. However, with a constant gradient descent step size, this spike only carries information from the linear component of the target function and therefore learning non-linear components is impossible. We show that with a learning rate that grows with the sample size, such training in fact introduces multiple rank-one components, each corresponding to a specific polynomial feature. We further prove that the limiting large-dimensional and large sample training and test errors of the updated neural networks are fully characterized by these spikes. By precisely analyzing the improvement in the loss, we demonstrate that these non-linear features can enhance learning.
Authors: Jalil Chavez-Galaviz, Jianwen Li, Ajinkya Chaudhary, Nina Mahmoudian
Station keeping is an essential maneuver for Autonomous Surface Vehicles (ASVs), mainly when used in confined spaces, to carry out surveys that require the ASV to keep its position or in collaboration with other vehicles where the relative position has an impact over the mission. However, this maneuver can become challenging for classic feedback controllers due to the need for an accurate model of the ASV dynamics and the environmental disturbances. This work proposes a Model Predictive Controller using Neural Network Simulation Error Minimization (NNSEM-MPC) to accurately predict the dynamics of the ASV under wind disturbances. The performance of the proposed scheme under wind disturbances is tested and compared against other controllers in simulation, using the Robotics Operating System (ROS) and the multipurpose simulation environment Gazebo. A set of six tests were conducted by combining two wind speeds (3 m/s and 6 m/s) and three wind directions (0$^\circ$, 90$^\circ$, and 180$^\circ$). The simulation results clearly show the advantage of the NNSEM-MPC over the following methods: backstepping controller, sliding mode controller, simplified dynamics MPC (SD-MPC), neural ordinary differential equation MPC (NODE-MPC), and knowledge-based NODE MPC (KNODE-MPC). The proposed NNSEM-MPC approach performs better than the rest in 4 out of the 6 test conditions, and it is the second best in the 2 remaining test cases, reducing the mean position and heading error by at least 31\% and 46\% respectively across all the test cases. In terms of execution speed, the proposed NNSEM-MPC is at least 36\% faster than the rest of the MPC controllers. The field experiments on two different ASV platforms showed that ASVs can effectively keep the station utilizing the proposed method, with a position error as low as $1.68$ m and a heading error as low as $6.14^{\circ}$ within time windows of at least $150$s.
Authors: Kushagra Pandey, Maja Rudolph, Stephan Mandt
Diffusion models suffer from slow sample generation at inference time. Therefore, developing a principled framework for fast deterministic/stochastic sampling for a broader class of diffusion models is a promising direction. We propose two complementary frameworks for accelerating sample generation in pre-trained models: Conjugate Integrators and Splitting Integrators. Conjugate integrators generalize DDIM, mapping the reverse diffusion dynamics to a more amenable space for sampling. In contrast, splitting-based integrators, commonly used in molecular dynamics, reduce the numerical simulation error by cleverly alternating between numerical updates involving the data and auxiliary variables. After extensively studying these methods empirically and theoretically, we present a hybrid method that leads to the best-reported performance for diffusion models in augmented spaces. Applied to Phase Space Langevin Diffusion [Pandey & Mandt, 2023] on CIFAR-10, our deterministic and stochastic samplers achieve FID scores of 2.11 and 2.36 in only 100 network function evaluations (NFE) as compared to 2.57 and 2.63 for the best-performing baselines, respectively. Our code and model checkpoints will be made publicly available at \url{https://github.com/mandt-lab/PSLD}.
Authors: Julia Werner, Christoph Gerum, Moritz Reiber, Jörg Nick, Oliver Bringmann
This paper presents a method to efficiently classify the gastroenterologic section of images derived from Video Capsule Endoscopy (VCE) studies by exploring the combination of a Convolutional Neural Network (CNN) for classification with the time-series analysis properties of a Hidden Markov Model (HMM). It is demonstrated that successive time-series analysis identifies and corrects errors in the CNN output. Our approach achieves an accuracy of $98.04\%$ on the Rhode Island (RI) Gastroenterology dataset. This allows for precise localization within the gastrointestinal (GI) tract while requiring only approximately 1M parameters and thus, provides a method suitable for low power devices
Authors: Ajay Sridhar, Dhruv Shah, Catherine Glossop, Sergey Levine
Robotic learning for navigation in unfamiliar environments needs to provide policies for both task-oriented navigation (i.e., reaching a goal that the robot has located), and task-agnostic exploration (i.e., searching for a goal in a novel setting). Typically, these roles are handled by separate models, for example by using subgoal proposals, planning, or separate navigation strategies. In this paper, we describe how we can train a single unified diffusion policy to handle both goal-directed navigation and goal-agnostic exploration, with the latter providing the ability to search novel environments, and the former providing the ability to reach a user-specified goal once it has been located. We show that this unified policy results in better overall performance when navigating to visually indicated goals in novel environments, as compared to approaches that use subgoal proposals from generative models, or prior methods based on latent variable models. We instantiate our method by using a large-scale Transformer-based policy trained on data from multiple ground robots, with a diffusion model decoder to flexibly handle both goal-conditioned and goal-agnostic navigation. Our experiments, conducted on a real-world mobile robot platform, show effective navigation in unseen environments in comparison with five alternative methods, and demonstrate significant improvements in performance and lower collision rates, despite utilizing smaller models than state-of-the-art approaches. For more videos, code, and pre-trained model checkpoints, see https://general-navigation-models.github.io/nomad/
Authors: Noémie Jaquier, Leonel Rozo, Tamim Asfour
In the realm of robotics, numerous downstream robotics tasks leverage machine learning methods for processing, modeling, or synthesizing data. Often, this data comprises variables that inherently carry geometric constraints, such as the unit-norm condition of quaternions representing rigid-body orientations or the positive definiteness of stiffness and manipulability ellipsoids. Handling such geometric constraints effectively requires the incorporation of tools from differential geometry into the formulation of machine learning methods. In this context, Riemannian manifolds emerge as a powerful mathematical framework to handle such geometric constraints. Nevertheless, their recent adoption in robot learning has been largely characterized by a mathematically-flawed simplification, hereinafter referred to as the ``single tangent space fallacy". This approach involves merely projecting the data of interest onto a single tangent (Euclidean) space, over which an off-the-shelf learning algorithm is applied. This paper provides a theoretical elucidation of various misconceptions surrounding this approach and offers experimental evidence of its shortcomings. Finally, it presents valuable insights to promote best practices when employing Riemannian geometry within robot learning applications.
Authors: Elaheh Jafarigol, Theodore Trafalis
For over two decades, detecting rare events has been a challenging task among researchers in the data mining and machine learning domain. Real-life problems inspire researchers to navigate and further improve data processing and algorithmic approaches to achieve effective and computationally efficient methods for imbalanced learning. In this paper, we have collected and reviewed 258 peer-reviewed papers from archival journals and conference papers in an attempt to provide an in-depth review of various approaches in imbalanced learning from technical and application perspectives. This work aims to provide a structured review of methods used to address the problem of imbalanced data in various domains and create a general guideline for researchers in academia or industry who want to dive into the broad field of machine learning using large-scale imbalanced data.
Authors: Jannik Deuschel, Caleb N. Ellington, Benjamin J. Lengerich, Yingtao Luo, Pascal Friederich, Eric P. Xing
Interpretable policy learning seeks to estimate intelligible decision policies from observed actions; however, existing models fall short by forcing a tradeoff between accuracy and interpretability. This tradeoff limits data-driven interpretations of human decision-making process. e.g. to audit medical decisions for biases and suboptimal practices, we require models of decision processes which provide concise descriptions of complex behaviors. Fundamentally, existing approaches are burdened by this tradeoff because they represent the underlying decision process as a universal policy, when in fact human decisions are dynamic and can change drastically with contextual information. Thus, we propose Contextualized Policy Recovery (CPR), which re-frames the problem of modeling complex decision processes as a multi-task learning problem in which complex decision policies are comprised of context-specific policies. CPR models each context-specific policy as a linear observation-to-action mapping, and generates new decision models $\textit{on-demand}$ as contexts are updated with new observations. CPR is compatible with fully offline and partially observable decision environments, and can be tailored to incorporate any recurrent black-box model or interpretable decision model. We assess CPR through studies on simulated and real data, achieving state-of-the-art performance on the canonical tasks of predicting antibiotic prescription in intensive care units ($+22\%$ AUROC vs. previous SOTA) and predicting MRI prescription for Alzheimer's patients ($+7.7\%$ AUROC vs. previous SOTA). With this improvement in predictive performance, CPR closes the accuracy gap between interpretable and black-box methods for policy learning, allowing high-resolution exploration and analysis of context-specific decision models.
Authors: William Merrill, Ashish Sabharwal
Recent theoretical work has identified surprisingly simple reasoning problems, such as checking if two nodes in a graph are connected or simulating finite-state machines, that are provably unsolvable by standard transformers that answer immediately after reading their input. However, in practice, transformers' reasoning can be improved by allowing them to use a "chain of thought" or "scratchpad", i.e., generate and condition on a sequence of intermediate tokens before answering. Motivated by this, we ask: Does such intermediate generation fundamentally extend the computational power of a decoder-only transformer? We show that the answer is yes, but the amount of increase depends crucially on the amount of intermediate generation. For instance, we find that transformer decoders with a logarithmic number of decoding steps (w.r.t. the input length) push the limits of standard transformers only slightly, while a linear number of decoding steps adds a clear new ability (under standard complexity conjectures): recognizing all regular languages. Our results also imply that linear steps keep transformer decoders within context-sensitive languages, and polynomial steps make them recognize exactly the class of polynomial-time solvable problems -- the first exact characterization of a type of transformers in terms of standard complexity classes. Together, our results provide a nuanced framework for understanding how the length of a transformer's chain of thought or scratchpad impacts its reasoning power.
Authors: M. Rostami, H. Moradian, S. S. Kia
This paper proposes a set of novel optimization algorithms for solving a class of convex optimization problems with time-varying streaming cost function. We develop an approach to track the optimal solution with a bounded error. Unlike the existing results, our algorithm is executed only by using the first-order derivatives of the cost function which makes it computationally efficient for optimization with time-varying cost function. We compare our algorithms to the gradient descent algorithm and show why gradient descent is not an effective solution for optimization problems with time-varying cost. Several examples including solving a model predictive control problem cast as a convex optimization problem with a streaming time-varying cost function demonstrate our results.
Authors: Ziyue Zou, Pratyush Tiwary
In this study, we present a graph neural network-based learning approach using an autoencoder setup to derive low-dimensional variables from features observed in experimental crystal structures. These variables are then biased in enhanced sampling to observe state-to-state transitions and reliable thermodynamic weights. Our approach uses simple convolution and pooling methods. To verify the effectiveness of our protocol, we examined the nucleation of various allotropes and polymorphs of iron and glycine from their molten states. Our graph latent variables when biased in well-tempered metadynamics consistently show transitions between states and achieve accurate free energy calculations in agreement with experiments, both of which are indicators of dependable sampling. This underscores the strength and promise of our graph neural net variables for improved sampling. The protocol shown here should be applicable for other systems and with other sampling methods.
Authors: Adyasha Maharana, Prateek Yadav, Mohit Bansal
Analytical theories suggest that higher-quality data can lead to lower test errors in models trained on a fixed data budget. Moreover, a model can be trained on a lower compute budget without compromising performance if a dataset can be stripped of its redundancies. Coreset selection (or data pruning) seeks to select a subset of the training data so as to maximize the performance of models trained on this subset, also referred to as coreset. There are two dominant approaches: (1) geometry-based data selection for maximizing data diversity in the coreset, and (2) functions that assign difficulty scores to samples based on training dynamics. Optimizing for data diversity leads to a coreset that is biased towards easier samples, whereas, selection by difficulty ranking omits easy samples that are necessary for the training of deep learning models. This demonstrates that data diversity and importance scores are two complementary factors that need to be jointly considered during coreset selection. We represent a dataset as an undirected graph and propose a novel pruning algorithm, D2 Pruning, that uses forward and reverse message passing over this dataset graph for coreset selection. D2 Pruning updates the difficulty scores of each example by incorporating the difficulty of its neighboring examples in the dataset graph. Then, these updated difficulty scores direct a graph-based sampling method to select a coreset that encapsulates both diverse and difficult regions of the dataset space. We evaluate supervised and self-supervised versions of our method on various vision and language datasets. Results show that D2 Pruning improves coreset selection over previous state-of-the-art methods for up to 70% pruning rates. Additionally, we find that using D2 Pruning for filtering large multimodal datasets leads to increased diversity in the dataset and improved generalization of pretrained models.
Authors: Ravit Sharma, Wojciech Romaszkan, Feiqian Zhu, Puneet Gupta
Researchers have long touted a vision of the future enabled by a proliferation of internet-of-things devices, including smart sensors, homes, and cities. Increasingly, embedding intelligence in such devices involves the use of deep neural networks. However, their storage and processing requirements make them prohibitive for cheap, off-the-shelf platforms. Overcoming those requirements is necessary for enabling widely-applicable smart devices. While many ways of making models smaller and more efficient have been developed, there is a lack of understanding of which ones are best suited for particular scenarios. More importantly for edge platforms, those choices cannot be analyzed in isolation from cost and user experience. In this work, we holistically explore how quantization, model scaling, and multi-modality interact with system components such as memory, sensors, and processors. We perform this hardware/software co-design from the cost, latency, and user-experience perspective, and develop a set of guidelines for optimal system design and model deployment for the most cost-constrained platforms. We demonstrate our approach using an end-to-end, on-device, biometric user authentication system using a $20 ESP-EYE board.
Authors: Md Mahbubur Rahman, Ira Ceka, Chengzhi Mao, Saikat Chakraborty, Baishakhi Ray, Wei Le
Deep learning vulnerability detection has shown promising results in recent years. However, an important challenge that still blocks it from being very useful in practice is that the model is not robust under perturbation and it cannot generalize well over the out-of-distribution (OOD) data, e.g., applying a trained model to unseen projects in real world. We hypothesize that this is because the model learned non-robust features, e.g., variable names, that have spurious correlations with labels. When the perturbed and OOD datasets no longer have the same spurious features, the model prediction fails. To address the challenge, in this paper, we introduced causality into deep learning vulnerability detection. Our approach CausalVul consists of two phases. First, we designed novel perturbations to discover spurious features that the model may use to make predictions. Second, we applied the causal learning algorithms, specifically, do-calculus, on top of existing deep learning models to systematically remove the use of spurious features and thus promote causal based prediction. Our results show that CausalVul consistently improved the model accuracy, robustness and OOD performance for all the state-of-the-art models and datasets we experimented. To the best of our knowledge, this is the first work that introduces do calculus based causal learning to software engineering models and shows it's indeed useful for improving the model accuracy, robustness and generalization. Our replication package is located at https://figshare.com/s/0ffda320dcb96c249ef2.
Authors: Abdullah Hayajneh, Erchin Serpedin, Mohammad Shaqfeh, Graeme Glass, Mitchell A. Stotland
A major obstacle when attempting to train a machine learning system to evaluate facial clefts is the scarcity of large datasets of high-quality, ethics board-approved patient images. In response, we have built a deep learning-based cleft lip generator designed to produce an almost unlimited number of artificial images exhibiting high-fidelity facsimiles of cleft lip with wide variation. We undertook a transfer learning protocol testing different versions of StyleGAN-ADA (a generative adversarial network image generator incorporating adaptive data augmentation (ADA)) as the base model. Training images depicting a variety of cleft deformities were pre-processed to adjust for rotation, scaling, color adjustment and background blurring. The ADA modification of the primary algorithm permitted construction of our new generative model while requiring input of a relatively small number of training images. Adversarial training was carried out using 514 unique frontal photographs of cleft-affected faces to adapt a pre-trained model based on 70,000 normal faces. The Frechet Inception Distance (FID) was used to measure the similarity of the newly generated facial images to the cleft training dataset, while Perceptual Path Length (PPL) and the novel Divergence Index of Severity Histograms (DISH) measures were also used to assess the performance of the image generator that we dub CleftGAN. We found that StyleGAN3 with translation invariance (StyleGAN3-t) performed optimally as a base model. Generated images achieved a low FID reflecting a close similarity to our training input dataset of genuine cleft images. Low PPL and DISH measures reflected a smooth and semantically valid interpolation of images through the transfer learning process and a similar distribution of severity in the training and generated images, respectively.
Authors: Nazanin Nezami, Hadis Anahideh
Surrogate Optimization (SO) algorithms have shown promise for optimizing expensive black-box functions. However, their performance is heavily influenced by hyperparameters related to sampling and surrogate fitting, which poses a challenge to their widespread adoption. We investigate the impact of hyperparameters on various SO algorithms and propose a Hyperparameter Adaptive Search for SO (HASSO) approach. HASSO is not a hyperparameter tuning algorithm, but a generic self-adjusting SO algorithm that dynamically tunes its own hyperparameters while concurrently optimizing the primary objective function, without requiring additional evaluations. The aim is to improve the accessibility, effectiveness, and convergence speed of SO algorithms for practitioners. Our approach identifies and modifies the most influential hyperparameters specific to each problem and SO approach, reducing the need for manual tuning without significantly increasing the computational burden. Experimental results demonstrate the effectiveness of HASSO in enhancing the performance of various SO algorithms across different global optimization test problems.
Authors: Xianghao Kong, Ollie Liu, Han Li, Dani Yogatama, Greg Ver Steeg
Denoising diffusion models enable conditional generation and density modeling of complex relationships like images and text. However, the nature of the learned relationships is opaque making it difficult to understand precisely what relationships between words and parts of an image are captured, or to predict the effect of an intervention. We illuminate the fine-grained relationships learned by diffusion models by noticing a precise relationship between diffusion and information decomposition. Exact expressions for mutual information and conditional mutual information can be written in terms of the denoising model. Furthermore, pointwise estimates can be easily estimated as well, allowing us to ask questions about the relationships between specific images and captions. Decomposing information even further to understand which variables in a high-dimensional space carry information is a long-standing problem. For diffusion models, we show that a natural non-negative decomposition of mutual information emerges, allowing us to quantify informative relationships between words and pixels in an image. We exploit these new relations to measure the compositional understanding of diffusion models, to do unsupervised localization of objects in images, and to measure effects when selectively editing images through prompt interventions.
Authors: Zohair Shafi, Benjamin A. Miller, Tina Eliassi-Rad, Rajmonda S. Caceres
Machine learning (ML) approaches are increasingly being used to accelerate combinatorial optimization (CO) problems. We look specifically at the Set Cover Problem (SCP) and propose Graph-SCP, a graph neural network method that can augment existing optimization solvers by learning to identify a much smaller sub-problem that contains the solution space. We evaluate the performance of Graph-SCP on synthetic weighted and unweighted SCP instances with diverse problem characteristics and complexities, and on instances from the OR Library, a canonical benchmark for SCP. We show that Graph-SCP reduces the problem size by 30-70% and achieves run time speedups up to~25x when compared to commercial solvers (Gurobi). Given a desired optimality threshold, Graph-SCP will improve upon it or even achieve 100% optimality. This is in contrast to fast greedy solutions that significantly compromise solution quality to achieve guaranteed polynomial run time. Graph-SCP can generalize to larger problem sizes and can be used with other conventional or ML-augmented CO solvers to lead to potential additional run time improvement.
Authors: Zohair Shafi. Benjamin A. Miller, Ayan Chatterjee, Tina Eliassi-Rad, Rajmonda S. Caceres
Recent advances in machine learning (ML) have shown promise in aiding and accelerating classical combinatorial optimization algorithms. ML-based speed ups that aim to learn in an end to end manner (i.e., directly output the solution) tend to trade off run time with solution quality. Therefore, solutions that are able to accelerate existing solvers while maintaining their performance guarantees, are of great interest. We consider an APX-hard problem, where an adversary aims to attack shortest paths in a graph by removing the minimum number of edges. We propose the GRASP algorithm: Graph Attention Accelerated Shortest Path Attack, an ML aided optimization algorithm that achieves run times up to 10x faster, while maintaining the quality of solution generated. GRASP uses a graph attention network to identify a smaller subgraph containing the combinatorial solution, thus effectively reducing the input problem size. Additionally, we demonstrate how careful representation of the input graph, including node features that correlate well with the optimization task, can highlight important structure in the optimization solution.
Authors: Hwajong Lee, Chan Kim, Seong-Woo Kim
A flow control system is a critical concept for increasing the production capacity of manufacturing systems. To solve the scheduling optimization problem related to the flow control with the aim of improving productivity, existing methods depend on a heuristic design by domain human experts. Therefore, the methods require correction, monitoring, and verification by using real equipment. As system designs increase in complexity, the monitoring time increases, which decreases the probability of arriving at the optimal design. As an alternative approach to the heuristic design of flow control systems, the use of deep reinforcement learning to solve the scheduling optimization problem has been considered. Although the existing research on reinforcement learning has yielded excellent performance in some areas, the applicability of the results to actual FAB such as display and semiconductor manufacturing processes is not evident so far. To this end, we propose a method to implement a physical simulation environment and devise a feasible flow control system design using a transfer robot in display manufacturing through reinforcement learning. We present a model and parameter setting to build a virtual environment for different display transfer robots, and training methods of reinforcement learning on the environment to obtain an optimal scheduling of glass flow control systems. Its feasibility was verified by using different types of robots used in the actual process.
Authors: Luyao Guo, Sulaiman A. Alghunaim, Kun Yuan, Laurent Condat, Jinde Cao
Distributed optimization methods with random communication skips are gaining increasing attention due to their proven benefits in accelerating communication complexity. Nevertheless, existing research mainly focuses on centralized communication protocols for strongly convex deterministic settings. In this work, we provide a decentralized optimization method called RandCom, which incorporates probabilistic local updates. We analyze the performance of RandCom in stochastic non-convex, convex, and strongly convex settings and demonstrate its ability to asymptotically reduce communication overhead by the probability of communication. Additionally, we prove that RandCom achieves linear speedup as the number of nodes increases. In stochastic strongly convex settings, we further prove that RandCom can achieve linear speedup with network-independent stepsizes. Moreover, we apply RandCom to federated learning and provide positive results concerning the potential for achieving linear speedup and the suitability of the probabilistic local update approach for non-convex settings.
Authors: Fu Luo, Xi Lin, Fei Liu, Qingfu Zhang, Zhenkun Wang
Neural combinatorial optimization (NCO) is a promising learning-based approach for solving challenging combinatorial optimization problems without specialized algorithm design by experts. However, most constructive NCO methods cannot solve problems with large-scale instance sizes, which significantly diminishes their usefulness for real-world applications. In this work, we propose a novel Light Encoder and Heavy Decoder (LEHD) model with a strong generalization ability to address this critical issue. The LEHD model can learn to dynamically capture the relationships between all available nodes of varying sizes, which is beneficial for model generalization to problems of various scales. Moreover, we develop a data-efficient training scheme and a flexible solution construction mechanism for the proposed LEHD model. By training on small-scale problem instances, the LEHD model can generate nearly optimal solutions for the Travelling Salesman Problem (TSP) and the Capacitated Vehicle Routing Problem (CVRP) with up to 1000 nodes, and also generalizes well to solve real-world TSPLib and CVRPLib problems. These results confirm our proposed LEHD model can significantly improve the state-of-the-art performance for constructive NCO. The code is available at https://github.com/CIAM-Group/NCO_code/tree/main/single_objective/LEHD.
Authors: Wensheng Lin, Yuna Yan, Lixin Li, Zhu Han, Tad Matsumoto
This letter proposes a novel relaying framework, semantic-forward (SF), for cooperative communications towards the sixth-generation (6G) wireless networks. The SF relay extracts and transmits the semantic features, which reduces forwarding payload, and also improves the network robustness against intra-link errors. Based on the theoretical basis for cooperative communications with side information and the turbo principle, we design a joint source-channel coding algorithm to iteratively exchange the extrinsic information for enhancing the decoding gains at the destination. Surprisingly, simulation results indicate that even in bad channel conditions, SF relaying can still effectively improve the recovered information quality.
Authors: Chen Zhao, Kuan-Jui Su, Chong Wu, Xuewei Cao, Qiuying Sha, Wu Li, Zhe Luo, Tian Qin, Chuan Qiu, Lan Juan Zhao, Anqi Liu, Lindong Jiang, Xiao Zhang, Hui Shen, Weihua Zhou, Hong-Wen Deng
Background: Missing data is a common challenge in mass spectrometry-based metabolomics, which can lead to biased and incomplete analyses. The integration of whole-genome sequencing (WGS) data with metabolomics data has emerged as a promising approach to enhance the accuracy of data imputation in metabolomics studies. Method: In this study, we propose a novel method that leverages the information from WGS data and reference metabolites to impute unknown metabolites. Our approach utilizes a multi-view variational autoencoder to jointly model the burden score, polygenetic risk score (PGS), and linkage disequilibrium (LD) pruned single nucleotide polymorphisms (SNPs) for feature extraction and missing metabolomics data imputation. By learning the latent representations of both omics data, our method can effectively impute missing metabolomics values based on genomic information. Results: We evaluate the performance of our method on empirical metabolomics datasets with missing values and demonstrate its superiority compared to conventional imputation techniques. Using 35 template metabolites derived burden scores, PGS and LD-pruned SNPs, the proposed methods achieved r2-scores > 0.01 for 71.55% of metabolites. Conclusion: The integration of WGS data in metabolomics imputation not only improves data completeness but also enhances downstream analyses, paving the way for more comprehensive and accurate investigations of metabolic pathways and disease associations. Our findings offer valuable insights into the potential benefits of utilizing WGS data for metabolomics data imputation and underscore the importance of leveraging multi-modal data integration in precision medicine research.
Authors: Lapo Frati, Neil Traft, Jeff Clune, Nick Cheney
This work identifies a simple pre-training mechanism that leads to representations exhibiting better continual and transfer learning. This mechanism -- the repeated resetting of weights in the last layer, which we nickname "zapping" -- was originally designed for a meta-continual-learning procedure, yet we show it is surprisingly applicable in many settings beyond both meta-learning and continual learning. In our experiments, we wish to transfer a pre-trained image classifier to a new set of classes, in a few shots. We show that our zapping procedure results in improved transfer accuracy and/or more rapid adaptation in both standard fine-tuning and continual learning settings, while being simple to implement and computationally efficient. In many cases, we achieve performance on par with state of the art meta-learning without needing the expensive higher-order gradients, by using a combination of zapping and sequential learning. An intuitive explanation for the effectiveness of this zapping procedure is that representations trained with repeated zapping learn features that are capable of rapidly adapting to newly initialized classifiers. Such an approach may be considered a computationally cheaper type of, or alternative to, meta-learning rapidly adaptable features with higher-order gradients. This adds to recent work on the usefulness of resetting neural network parameters during training, and invites further investigation of this mechanism.
Authors: Yite Wang, Jiahao Su, Hanlin Lu, Cong Xie, Tianyi Liu, Jianbo Yuan, Haibin Lin, Ruoyu Sun, Hongxia Yang
Scaling of deep neural networks, especially Transformers, is pivotal for their surging performance and has further led to the emergence of sophisticated reasoning capabilities in foundation models. Such scaling generally requires training large models from scratch with random initialization, failing to leverage the knowledge acquired by their smaller counterparts, which are already resource-intensive to obtain. To tackle this inefficiency, we present $\textbf{L}$ossl$\textbf{E}$ss $\textbf{MO}$del Expansio$\textbf{N}$ (LEMON), a recipe to initialize scaled models using the weights of their smaller but pre-trained counterparts. This is followed by model training with an optimized learning rate scheduler tailored explicitly for the scaled models, substantially reducing the training time compared to training from scratch. Notably, LEMON is versatile, ensuring compatibility with various network structures, including models like Vision Transformers and BERT. Our empirical results demonstrate that LEMON reduces computational costs by 56.7% for Vision Transformers and 33.2% for BERT when compared to training from scratch.
Authors: Wei Ao, Vishnu Naresh Boddeti
Secure inference of deep convolutional neural networks (CNNs) under RNS-CKKS involves polynomial approximation of unsupported non-linear activation functions. However, existing approaches have three main limitations: 1) Inflexibility: The polynomial approximation and associated homomorphic evaluation architecture are customized manually for each CNN architecture and do not generalize to other networks. 2) Suboptimal Approximation: Each activation function is approximated instead of the function represented by the CNN. 3) Restricted Design: Either high-degree or low-degree polynomial approximations are used. The former retains high accuracy but slows down inference due to bootstrapping operations, while the latter accelerates ciphertext inference but compromises accuracy. To address these limitations, we present AutoFHE, which automatically adapts standard CNNs for secure inference under RNS-CKKS. The key idea is to adopt layerwise mixed-degree polynomial activation functions, which are optimized jointly with the homomorphic evaluation architecture in terms of the placement of bootstrapping operations. The problem is modeled within a multi-objective optimization framework to maximize accuracy and minimize the number of bootstrapping operations. AutoFHE can be applied flexibly on any CNN architecture, and it provides diverse solutions that span the trade-off between accuracy and latency. Experimental evaluation over RNS-CKKS encrypted CIFAR datasets shows that AutoFHE accelerates secure inference by $1.32\times$ to $1.8\times$ compared to methods employing high-degree polynomials. It also improves accuracy by up to 2.56% compared to methods using low-degree polynomials. Lastly, AutoFHE accelerates inference and improves accuracy by $103\times$ and 3.46%, respectively, compared to CNNs under TFHE.
Authors: Jihye Choi, Shruti Tople, Varun Chandrasekaran, Somesh Jha
Membership Inference Attacks (MIAs) aim to identify specific data samples within the private training dataset of machine learning models, leading to serious privacy violations and other sophisticated threats. Many practical black-box MIAs require query access to the data distribution (the same distribution where the private data is drawn) to train shadow models. By doing so, the adversary obtains models trained "with" or "without" samples drawn from the distribution, and analyzes the characteristics of the samples under consideration. The adversary is often required to train more than hundreds of shadow models to extract the signals needed for MIAs; this becomes the computational overhead of MIAs. In this paper, we propose that by strategically choosing the samples, MI adversaries can maximize their attack success while minimizing the number of shadow models. First, our motivational experiments suggest memorization as the key property explaining disparate sample vulnerability to MIAs. We formalize this through a theoretical bound that connects MI advantage with memorization. Second, we show sample complexity bounds that connect the number of shadow models needed for MIAs with memorization. Lastly, we confirm our theoretical arguments with comprehensive experiments; by utilizing samples with high memorization scores, the adversary can (a) significantly improve its efficacy regardless of the MIA used, and (b) reduce the number of shadow models by nearly two orders of magnitude compared to state-of-the-art approaches.
Authors: Namiko Matsumoto, Arya Mazumdar
In 1-bit compressed sensing, the aim is to estimate a $k$-sparse unit vector $x\in S^{n-1}$ within an $\epsilon$ error (in $\ell_2$) from minimal number of linear measurements that are quantized to just their signs, i.e., from measurements of the form $y = \mathrm{Sign}(\langle a, x\rangle).$ In this paper, we study a noisy version where a fraction of the measurements can be flipped, potentially by an adversary. In particular, we analyze the Binary Iterative Hard Thresholding (BIHT) algorithm, a proximal gradient descent on a properly defined loss function used for 1-bit compressed sensing, in this noisy setting. It is known from recent results that, with $\tilde{O}(\frac{k}{\epsilon})$ noiseless measurements, BIHT provides an estimate within $\epsilon$ error. This result is optimal and universal, meaning one set of measurements work for all sparse vectors. In this paper, we show that BIHT also provides better results than all known methods for the noisy setting. We show that when up to $\tau$-fraction of the sign measurements are incorrect (adversarial error), with the same number of measurements as before, BIHT agnostically provides an estimate of $x$ within an $\tilde{O}(\epsilon+\tau)$ error, maintaining the universality of measurements. This establishes stability of iterative hard thresholding in the presence of measurement error. To obtain the result, we use the restricted approximate invertibility of Gaussian matrices, as well as a tight analysis of the high-dimensional geometry of the adversarially corrupted measurements.
Authors: Artur Back de Luca, Kimon Fountoulakis, Shenghao Yang
The growing interest in machine learning problems over graphs with additional node information such as texts, images, or labels has popularized methods that require the costly operation of processing the entire graph. Yet, little effort has been made to the development of fast local methods (i.e. without accessing the entire graph) that extract useful information from such data. To that end, we propose a study of local graph clustering using noisy node labels as a proxy for additional node information. In this setting, nodes receive initial binary labels based on cluster affiliation: 1 if they belong to the target cluster and 0 otherwise. Subsequently, a fraction of these labels is flipped. We investigate the benefits of incorporating noisy labels for local graph clustering. By constructing a weighted graph with such labels, we study the performance of graph diffusion-based local clustering method on both the original and the weighted graphs. From a theoretical perspective, we consider recovering an unknown target cluster with a single seed node in a random graph with independent noisy node labels. We provide sufficient conditions on the label noise under which, with high probability, using diffusion in the weighted graph yields a more accurate recovery of the target cluster. This approach proves more effective than using the given labels alone or using diffusion in the label-free original graph. Empirically, we show that reliable node labels can be obtained with just a few samples from an attributed graph. Moreover, utilizing these labels via diffusion in the weighted graph leads to significantly better local clustering performance across several real-world datasets, improving F1 scores by up to 13%.
Authors: Binghui Wu, Philipp Gysel, Dinil Mon Divakaran, Mohan Gurusamy
Recent research works have proposed machine learning models for classifying IoT devices connected to a network. However, there is still a practical challenge of not having all devices (and hence their traffic) available during the training of a model. This essentially means, during the operational phase, we need to classify new devices not seen during the training phase. To address this challenge, we propose ZEST -- a ZSL (zero-shot learning) framework based on self-attention for classifying both seen and unseen devices. ZEST consists of i) a self-attention based network feature extractor, termed SANE, for extracting latent space representations of IoT traffic, ii) a generative model that trains a decoder using latent features to generate pseudo data, and iii) a supervised model that is trained on the generated pseudo data for classifying devices. We carry out extensive experiments on real IoT traffic data; our experiments demonstrate i) ZEST achieves significant improvement (in terms of accuracy) over the baselines; ii) ZEST is able to better extract meaningful representations than LSTM which has been commonly used for modeling network traffic.
Authors: Zihao Xu, Xuan Tang, Yufei Shi, Jianfeng Zhang, Jian Yang, Mingsong Chen, Xian Wei
In continual learning, the learner learns multiple tasks in sequence, with data being acquired only once for each task. Catastrophic forgetting is a major challenge to continual learning. To reduce forgetting, some existing rehearsal-based methods use episodic memory to replay samples of previous tasks. However, in the process of knowledge integration when learning a new task, this strategy also suffers from catastrophic forgetting due to an imbalance between old and new knowledge. To address this problem, we propose a novel replay strategy called Manifold Expansion Replay (MaER). We argue that expanding the implicit manifold of the knowledge representation in the episodic memory helps to improve the robustness and expressiveness of the model. To this end, we propose a greedy strategy to keep increasing the diameter of the implicit manifold represented by the knowledge in the buffer during memory management. In addition, we introduce Wasserstein distance instead of cross entropy as distillation loss to preserve previous knowledge. With extensive experimental validation on MNIST, CIFAR10, CIFAR100, and TinyImageNet, we show that the proposed method significantly improves the accuracy in continual learning setup, outperforming the state of the arts.
Authors: Jinbo Song (1), Ruoran Huang (1), Xinyang Wang (1), Wei Huang (1), Qian Yu (1), Mingming Chen (1), Yafei Yao (1), Chaosheng Fan (1), Changping Peng (1), Zhangang Lin (1), Jinghe Hu (1), Jingping Shao (1) ((1) Marketing and Commercialization Center, JD.com)
Industrial systems such as recommender systems and online advertising, have been widely equipped with multi-stage architectures, which are divided into several cascaded modules, including matching, pre-ranking, ranking and re-ranking. As a critical bridge between matching and ranking, existing pre-ranking approaches mainly endure sample selection bias (SSB) problem owing to ignoring the entire-chain data dependence, resulting in sub-optimal performances. In this paper, we rethink pre-ranking system from the perspective of the entire sample space, and propose Entire-chain Cross-domain Models (ECM), which leverage samples from the whole cascaded stages to effectively alleviate SSB problem. Besides, we design a fine-grained neural structure named ECMM to further improve the pre-ranking accuracy. Specifically, we propose a cross-domain multi-tower neural network to comprehensively predict for each stage result, and introduce the sub-networking routing strategy with $L0$ regularization to reduce computational costs. Evaluations on real-world large-scale traffic logs demonstrate that our pre-ranking models outperform SOTA methods while time consumption is maintained within an acceptable level, which achieves better trade-off between efficiency and effectiveness.
Authors: Xiaoyang Song, Wenbo Sun, Maher Nouiehed, Raed Al Kontar, Judy Jin
Current techniques for Out-of-Distribution (OoD) detection predominantly rely on quantifying predictive uncertainty and incorporating model regularization during the training phase, using either real or synthetic OoD samples. However, methods that utilize real OoD samples lack exploration and are prone to overfit the OoD samples at hand. Whereas synthetic samples are often generated based on features extracted from training data, rendering them less effective when the training and OoD data are highly overlapped in the feature space. In this work, we propose a Wasserstein-score-based generative adversarial training scheme to enhance OoD detection accuracy, which, for the first time, performs data augmentation and exploration simultaneously under the supervision of limited OoD samples. Specifically, the generator explores OoD spaces and generates synthetic OoD samples using feedback from the discriminator, while the discriminator exploits both the observed and synthesized samples for OoD detection using a predefined Wasserstein score. We provide theoretical guarantees that the optimal solutions of our generative scheme are statistically achievable through adversarial training in empirical settings. We then demonstrate that the proposed method outperforms state-of-the-art techniques on various computer vision datasets and exhibits superior generalizability to unseen OoD data.
Authors: Jing Liu, Ruihao Gong, Xiuying Wei, Zhiwei Dong, Jianfei Cai, Bohan Zhuang
Large Language Models (LLMs) excel in NLP, but their demands hinder their widespread deployment. While Quantization-Aware Training (QAT) offers a solution, its extensive training costs make Post-Training Quantization (PTQ) a more practical approach for LLMs. In existing studies, activation outliers in particular channels are identified as the bottleneck to PTQ accuracy. They propose to transform the magnitudes from activations to weights, which however offers limited alleviation or suffers from unstable gradients, resulting in a severe performance drop at low-bitwidth. In this paper, we propose QLLM, an accurate and efficient low-bitwidth PTQ method designed for LLMs. QLLM introduces an adaptive channel reassembly technique that reallocates the magnitude of outliers to other channels, thereby mitigating their impact on the quantization range. This is achieved by channel disassembly and channel assembly, which first breaks down the outlier channels into several sub-channels to ensure a more balanced distribution of activation magnitudes. Then similar channels are merged to maintain the original channel number for efficiency. Additionally, an adaptive strategy is designed to autonomously determine the optimal number of sub-channels for channel disassembly. To further compensate for the performance loss caused by quantization, we propose an efficient tuning method that only learns a small number of low-rank weights while freezing the pre-trained quantized model. After training, these low-rank parameters can be fused into the frozen weights without affecting inference. Extensive experiments on LLaMA-1 and LLaMA-2 show that QLLM can obtain accurate quantized models efficiently. For example, QLLM quantizes the 4-bit LLaMA-2-70B within 10 hours on a single A100-80G GPU, outperforming the previous state-of-the-art method by 7.89% on the average accuracy across five zero-shot tasks.
Authors: Ivan Lee, Nan Jiang, Taylor Berg-Kirkpatrick
What is the relationship between model architecture and the ability to perform in-context learning? In this empirical study, we take the first steps towards answering this question. In particular, we evaluate fifteen model architectures across a suite of synthetic in-context learning tasks. The selected architectures represent a broad range of paradigms, including recurrent and convolution-based neural networks, transformers, and emerging attention alternatives. We discover that all considered architectures can perform in-context learning under certain conditions. However, contemporary architectures are found to be the best performing, especially as task complexity grows. Additionally, our follow-up experiments delve into various factors that influence in-context learning. We observe varied sensitivities among architectures with respect to hyperparameter settings. Our study of training dynamics reveals that certain architectures exhibit a smooth, progressive learning trajectory, while others demonstrate periods of stagnation followed by abrupt mastery of the task. Finally, and somewhat surprisingly, we find that several emerging attention alternatives are more robust in-context learners than transformers; since such approaches have constant-sized memory footprints at inference time, this result opens the future possibility of scaling up in-context learning to vastly larger numbers of in-context examples.
Authors: Jianchao Lu, Yuzhe Tian, Yang Zhang, Jiaqi Ge, Quan Z. Sheng, Xi Zheng
Brain-Computer Interfaces (BCIs) are a groundbreaking technology for interacting with external devices using brain signals. Despite advancements, electroencephalogram (EEG)-based Motor Imagery (MI) tasks face challenges like amplitude and phase variability, and complex spatial correlations, with a need for smaller model size and faster inference. This study introduces the LGL-BCI framework, employing a Geometric Deep Learning Framework for EEG processing in non-Euclidean metric spaces, particularly the Symmetric Positive Definite (SPD) Manifold space. LGL-BCI offers robust EEG data representation and captures spatial correlations. We propose an EEG channel selection solution via a feature decomposition algorithm to reduce SPD matrix dimensionality, with a lossless transformation boosting inference speed. Extensive experiments show LGL-BCI's superior accuracy and efficiency compared to current solutions, highlighting geometric deep learning's potential in MI-BCI applications. The efficiency, assessed on two public EEG datasets and two real-world EEG devices, significantly outperforms the state-of-the-art solution in accuracy ($82.54\%$ versus $62.22\%$) with fewer parameters (64.9M compared to 183.7M).
Authors: Shreyas Havaldar, Navodita Sharma, Shubhi Sareen, Karthikeyan Shanmugam, Aravindan Raghuveer
Learning from Label Proportions (LLP) is a learning problem where only aggregate level labels are available for groups of instances, called bags, during training, and the aim is to get the best performance at the instance-level on the test data. This setting arises in domains like advertising and medicine due to privacy considerations. We propose a novel algorithmic framework for this problem that iteratively performs two main steps. For the first step (Pseudo Labeling) in every iteration, we define a Gibbs distribution over binary instance labels that incorporates a) covariate information through the constraint that instances with similar covariates should have similar labels and b) the bag level aggregated label. We then use Belief Propagation (BP) to marginalize the Gibbs distribution to obtain pseudo labels. In the second step (Embedding Refinement), we use the pseudo labels to provide supervision for a learner that yields a better embedding. Further, we iterate on the two steps again by using the second step's embeddings as new covariates for the next iteration. In the final iteration, a classifier is trained using the pseudo labels. Our algorithm displays strong gains against several SOTA baselines (up to 15%) for the LLP Binary Classification problem on various dataset types - tabular and Image. We achieve these improvements with minimal computational overhead above standard supervised learning due to Belief Propagation, for large bag sizes, even for a million samples.
Authors: Yiqiang Yi, Xu Wan, Yatao Bian, Le Ou-Yang, Peilin Zhao
Predicting the docking between proteins and ligands is a crucial and challenging task for drug discovery. However, traditional docking methods mainly rely on scoring functions, and deep learning-based docking approaches usually neglect the 3D spatial information of proteins and ligands, as well as the graph-level features of ligands, which limits their performance. To address these limitations, we propose an equivariant transformer neural network for protein-ligand docking pose prediction. Our approach involves the fusion of ligand graph-level features by feature processing, followed by the learning of ligand and protein representations using our proposed TAMformer module. Additionally, we employ an iterative optimization approach based on the predicted distance matrix to generate refined ligand poses. The experimental results on real datasets show that our model can achieve state-of-the-art performance.
Authors: Haochen Li, Xin Zhou, Luu Anh Tuan, Chunyan Miao
Recently, contrastive learning has become a key component in fine-tuning code search models for software development efficiency and effectiveness. It pulls together positive code snippets while pushing negative samples away given search queries. Among contrastive learning, InfoNCE is the most widely used loss function due to its better performance. However, the following problems in negative samples of InfoNCE may deteriorate its representation learning: 1) The existence of false negative samples in large code corpora due to duplications. 2). The failure to explicitly differentiate between the potential relevance of negative samples. As an example, a bubble sorting algorithm example is less ``negative'' than a file saving function for the quick sorting algorithm query. In this paper, we tackle the above problems by proposing a simple yet effective Soft-InfoNCE loss that inserts weight terms into InfoNCE. In our proposed loss function, we apply three methods to estimate the weights of negative pairs and show that the vanilla InfoNCE loss is a special case of Soft-InfoNCE. Theoretically, we analyze the effects of Soft-InfoNCE on controlling the distribution of learnt code representations and on deducing a more precise mutual information estimation. We furthermore discuss the superiority of proposed loss functions with other design alternatives. Extensive experiments demonstrate the effectiveness of Soft-InfoNCE and weights estimation methods under state-of-the-art code search models on a large-scale public dataset consisting of six programming languages. Source code is available at \url{https://github.com/Alex-HaochenLi/Soft-InfoNCE}.
Authors: Xin Lyu, Avishay Tal, Hongxun Wu, Junzhao Yang
In his breakthrough paper, Raz showed that any parity learning algorithm requires either quadratic memory or an exponential number of samples [FOCS'16, JACM'19]. A line of work that followed extended this result to a large class of learning problems. Until recently, all these results considered learning in the streaming model, where each sample is drawn independently, and the learner is allowed a single pass over the stream of samples. Garg, Raz, and Tal [CCC'19] considered a stronger model, allowing multiple passes over the stream. In the $2$-pass model, they showed that learning parities of size $n$ requires either a memory of size $n^{1.5}$ or at least $2^{\sqrt{n}}$ samples. (Their result also generalizes to other learning problems.)
In this work, for any constant $q$, we prove tight memory-sample lower bounds for any parity learning algorithm that makes $q$ passes over the stream of samples. We show that such a learner requires either $\Omega(n^{2})$ memory size or at least $2^{\Omega(n)}$ samples. Beyond establishing a tight lower bound, this is the first non-trivial lower bound for $q$-pass learning for any $q\ge 3$. Similar to prior work, our results extend to any learning problem with many nearly-orthogonal concepts.
We complement the lower bound with an upper bound, showing that parity learning with $q$ passes can be done efficiently with $O(n^2/\log q)$ memory.
Authors: Junyu Gao, Xinhong Ma, Changsheng Xu
Despite the great progress of unsupervised domain adaptation (UDA) with the deep neural networks, current UDA models are opaque and cannot provide promising explanations, limiting their applications in the scenarios that require safe and controllable model decisions. At present, a surge of work focuses on designing deep interpretable methods with adequate data annotations and only a few methods consider the distributional shift problem. Most existing interpretable UDA methods are post-hoc ones, which cannot facilitate the model learning process for performance enhancement. In this paper, we propose an inherently interpretable method, named Transferable Conceptual Prototype Learning (TCPL), which could simultaneously interpret and improve the processes of knowledge transfer and decision-making in UDA. To achieve this goal, we design a hierarchically prototypical module that transfers categorical basic concepts from the source domain to the target domain and learns domain-shared prototypes for explaining the underlying reasoning process. With the learned transferable prototypes, a self-predictive consistent pseudo-label strategy that fuses confidence, predictions, and prototype information, is designed for selecting suitable target samples for pseudo annotations and gradually narrowing down the domain gap. Comprehensive experiments show that the proposed method can not only provide effective and intuitive explanations but also outperform previous state-of-the-arts.
Authors: Giorgio Piras, Maura Pintor, Ambra Demontis, Battista Biggio
Neural network pruning has shown to be an effective technique for reducing the network size, trading desirable properties like generalization and robustness to adversarial attacks for higher sparsity. Recent work has claimed that adversarial pruning methods can produce sparse networks while also preserving robustness to adversarial examples. In this work, we first re-evaluate three state-of-the-art adversarial pruning methods, showing that their robustness was indeed overestimated. We then compare pruned and dense versions of the same models, discovering that samples on thin ice, i.e., closer to the unpruned model's decision boundary, are typically misclassified after pruning. We conclude by discussing how this intuition may lead to designing more effective adversarial pruning methods in future work.
Authors: Md Mushfiqur Rahman, Fardin Ahsan Sakib, Fahim Faisal, Antonios Anastasopoulos
Choosing an appropriate tokenization scheme is often a bottleneck in low-resource cross-lingual transfer. To understand the downstream implications of text representation choices, we perform a comparative analysis on language models having diverse text representation modalities including 2 segmentation-based models (\texttt{BERT}, \texttt{mBERT}), 1 image-based model (\texttt{PIXEL}), and 1 character-level model (\texttt{CANINE}). First, we propose a scoring Language Quotient (LQ) metric capable of providing a weighted representation of both zero-shot and few-shot evaluation combined. Utilizing this metric, we perform experiments comprising 19 source languages and 133 target languages on three tasks (POS tagging, Dependency parsing, and NER). Our analysis reveals that image-based models excel in cross-lingual transfer when languages are closely related and share visually similar scripts. However, for tasks biased toward word meaning (POS, NER), segmentation-based models prove to be superior. Furthermore, in dependency parsing tasks where word relationships play a crucial role, models with their character-level focus, outperform others. Finally, we propose a recommendation scheme based on our findings to guide model selection according to task and language requirements.
Authors: Luca Barbieri, Stefano Savazzi, Sanaz Kianoush, Monica Nicoli, Luigi Serio
Federated Learning (FL) methods adopt efficient communication technologies to distribute machine learning tasks across edge devices, reducing the overhead in terms of data storage and computational complexity compared to centralized solutions. Rather than moving large data volumes from producers (sensors, machines) to energy-hungry data centers, raising environmental concerns due to resource demands, FL provides an alternative solution to mitigate the energy demands of several learning tasks while enabling new Artificial Intelligence of Things (AIoT) applications. This paper proposes a framework for real-time monitoring of the energy and carbon footprint impacts of FL systems. The carbon tracking tool is evaluated for consensus (fully decentralized) and classical FL policies. For the first time, we present a quantitative evaluation of different computationally and communication efficient FL methods from the perspectives of energy consumption and carbon equivalent emissions, suggesting also general guidelines for energy-efficient design. Results indicate that consensus-driven FL implementations should be preferred for limiting carbon emissions when the energy efficiency of the communication is low (i.e., < 25 Kbit/Joule). Besides, quantization and sparsification operations are shown to strike a balance between learning performances and energy consumption, leading to sustainable FL designs.
Authors: Jože M. Rožanec, Gašper Petelin, João Costa, Blaž Bertalanič, Gregor Cerar, Marko Guček, Gregor Papa, Dunja Mladenić
In many cases, a machine learning model must learn to correctly predict a few data points with particular values of interest in a broader range of data where many target values are zero. Zero-inflated data can be found in diverse scenarios, such as lumpy and intermittent demands, power consumption for home appliances being turned on and off, impurities measurement in distillation processes, and even airport shuttle demand prediction. The presence of zeroes affects the models' learning and may result in poor performance. Furthermore, zeroes also distort the metrics used to compute the model's prediction quality. This paper showcases two real-world use cases (home appliances classification and airport shuttle demand prediction) where a hierarchical model applied in the context of zero-inflated data leads to excellent results. In particular, for home appliances classification, the weighted average of Precision, Recall, F1, and AUC ROC was increased by 27%, 34%, 49%, and 27%, respectively. Furthermore, it is estimated that the proposed approach is also four times more energy efficient than the SOTA approach against which it was compared to. Two-fold models performed best in all cases when predicting airport shuttle demand, and the difference against other models has been proven to be statistically significant.
Authors: Jianfei Ma
Temporal difference learning (TD) is a foundational concept in reinforcement learning (RL), aimed at efficiently assessing a policy's value function. TD($\lambda$), a potent variant, incorporates a memory trace to distribute the prediction error into the historical context. However, this approach often neglects the significance of historical states and the relative importance of propagating the TD error, influenced by challenges such as visitation imbalance or outcome noise. To address this, we propose a novel TD algorithm named discerning TD learning (DTD), which allows flexible emphasis functions$-$predetermined or adapted during training$-$to allocate efforts effectively across states. We establish the convergence properties of our method within a specific class of emphasis functions and showcase its promising potential for adaptation to deep RL contexts. Empirical results underscore that employing a judicious emphasis function not only improves value estimation but also expedites learning across diverse scenarios.
Authors: Tobias Schimanski, Julia Bingler, Camilla Hyslop, Mathias Kraus, Markus Leippold
Public and private actors struggle to assess the vast amounts of information about sustainability commitments made by various institutions. To address this problem, we create a novel tool for automatically detecting corporate, national, and regional net zero and reduction targets in three steps. First, we introduce an expert-annotated data set with 3.5K text samples. Second, we train and release ClimateBERT-NetZero, a natural language classifier to detect whether a text contains a net zero or reduction target. Third, we showcase its analysis potential with two use cases: We first demonstrate how ClimateBERT-NetZero can be combined with conventional question-answering (Q&A) models to analyze the ambitions displayed in net zero and reduction targets. Furthermore, we employ the ClimateBERT-NetZero model on quarterly earning call transcripts and outline how communication patterns evolve over time. Our experiments demonstrate promising pathways for extracting and analyzing net zero and emission reduction targets at scale.
Authors: Jianfei Ma
Future sequence represents the outcome after executing the action into the environment. When driven by the information-theoretic concept of mutual information, it seeks maximally informative consequences. Explicit outcomes may vary across state, return, or trajectory serving different purposes such as credit assignment or imitation learning. However, the inherent nature of incorporating intrinsic motivation with reward maximization is often neglected. In this work, we propose a variational approach to jointly learn the necessary quantity for estimating the mutual information and the dynamics model, providing a general framework for incorporating different forms of outcomes of interest. Integrated into a policy iteration scheme, our approach guarantees convergence to the optimal policy. While we mainly focus on theoretical analysis, our approach opens the possibilities of leveraging intrinsic control with model learning to enhance sample efficiency and incorporate uncertainty of the environment into decision-making.
Authors: Gerard T. Schuster, Shihang Feng
We review four types of algorithms for physics-informed machine learning (PIML) inversion of geophysical data. The unifying equation is given by the joint objective function $\epsilon$:
\begin{eqnarray} \epsilon^{||-PIML}&=&\lambda_1 \overbrace{||{\bf W}^{ML}({\bf H}_{{\bf w}} {\bf d}^{obs}-{\bf m})||^2}^{NN} + \lambda_2 \overbrace{{||{\bf W}^{FWI}({\bf L} {\bf m}-{\bf d}^{obs})||^2}}^{FWI} ~+ \nonumber\\ \nonumber\\ && + ~~Regularizer, \label{PIML.eq120} \end{eqnarray}where the optimal model ${\bf m}^*$ and weights $\bf w^*$ minimize $\epsilon$. Here, The matrix weights are given by the boldface symbol $\bf W$, and full waveform inversion (FWI) is typically computed using a finite-difference solution of the wave equation, where $\bf L$ represents the forward modeling operation of the wave equation as a function of the model $\bf m$. Also, a fully-connected neural network (NN) is used to compute the model ${\bf H_w}{\bf d}^{obs} \approx \bf m$ from the observed input data ${\bf d}^{obs}$. The selection of weights $\lambda_i$ and the NN operations determine one of four different PIML algorithms.
PIML offers potential advantages over standard FWI through its enhanced ability to avoid local minima and the option to locally train the inversion operator, minimizing the requirement for extensive training data for global applicability. However, the effectiveness of PIML relies on the similarity between the test and trained data. Nevertheless, a possible strategy to overcome this limitation involves initial pretraining of a PIML architecture with data from a broader region, followed by fine-tuning for specific data-a method reminiscent of the way large language models are pretrained and adapted for various tasks.
Authors: Sepideh Mahabadi, Stojan Trajanovski
We study core-set construction algorithms for the task of Diversity Maximization under fairness/partition constraint. Given a set of points $P$ in a metric space partitioned into $m$ groups, and given $k_1,\ldots,k_m$, the goal of this problem is to pick $k_i$ points from each group $i$ such that the overall diversity of the $k=\sum_i k_i$ picked points is maximized. We consider two natural diversity measures: sum-of-pairwise distances and sum-of-nearest-neighbor distances, and show improved core-set construction algorithms with respect to these measures. More precisely, we show the first constant factor core-set w.r.t. sum-of-pairwise distances whose size is independent of the size of the dataset and the aspect ratio. Second, we show the first core-set w.r.t. the sum-of-nearest-neighbor distances. Finally, we run several experiments showing the effectiveness of our core-set approach. In particular, we apply constrained diversity maximization to summarize a set of timed messages that takes into account the messages' recency. Specifically, the summary should include more recent messages compared to older ones. This is a real task in one of the largest communication platforms, affecting the experience of hundreds of millions daily active users. By utilizing our core-set method for this task, we achieve a 100x speed-up while losing the diversity by only a few percent. Moreover, our approach allows us to improve the space usage of the algorithm in the streaming setting.
Authors: Zhendong Wang, Ioanna Miliou, Isak Samsten, Panagiotis Papapetrou
Among recent developments in time series forecasting methods, deep forecasting models have gained popularity as they can utilize hidden feature patterns in time series to improve forecasting performance. Nevertheless, the majority of current deep forecasting models are opaque, hence making it challenging to interpret the results. While counterfactual explanations have been extensively employed as a post-hoc approach for explaining classification models, their application to forecasting models still remains underexplored. In this paper, we formulate the novel problem of counterfactual generation for time series forecasting, and propose an algorithm, called ForecastCF, that solves the problem by applying gradient-based perturbations to the original time series. ForecastCF guides the perturbations by applying constraints to the forecasted values to obtain desired prediction outcomes. We experimentally evaluate ForecastCF using four state-of-the-art deep model architectures and compare to two baselines. Our results show that ForecastCF outperforms the baseline in terms of counterfactual validity and data manifold closeness. Overall, our findings suggest that ForecastCF can generate meaningful and relevant counterfactual explanations for various forecasting tasks.
Authors: Haiyang Liu, Chunjiang Zhu, Detian Zhang, Qing Li
Traffic flow prediction is one of the most fundamental tasks of intelligent transportation systems. The complex and dynamic spatial-temporal dependencies make the traffic flow prediction quite challenging. Although existing spatial-temporal graph neural networks hold prominent, they often encounter challenges such as (1) ignoring the fixed graph that limits the predictive performance of the model, (2) insufficiently capturing complex spatial-temporal dependencies simultaneously, and (3) lacking attention to spatial-temporal information at different time lengths. In this paper, we propose a Multi-Scale Spatial-Temporal Recurrent Network for traffic flow prediction, namely MSSTRN, which consists of two different recurrent neural networks: the single-step gate recurrent unit and the multi-step gate recurrent unit to fully capture the complex spatial-temporal information in the traffic data under different time windows. Moreover, we propose a spatial-temporal synchronous attention mechanism that integrates adaptive position graph convolutions into the self-attention mechanism to achieve synchronous capture of spatial-temporal dependencies. We conducted extensive experiments on four real traffic datasets and demonstrated that our model achieves the best prediction accuracy with non-trivial margins compared to all the twenty baseline methods.
Authors: Jingru Gan, Xinzhe Han, Shuhui Wang, Qingming Huang
Given an image and an associated textual question, the purpose of Knowledge-Based Visual Question Answering (KB-VQA) is to provide a correct answer to the question with the aid of external knowledge bases. Prior KB-VQA models are usually formulated as a retriever-classifier framework, where a pre-trained retriever extracts textual or visual information from knowledge graphs and then makes a prediction among the candidates. Despite promising progress, there are two drawbacks with existing models. Firstly, modeling question-answering as multi-class classification limits the answer space to a preset corpus and lacks the ability of flexible reasoning. Secondly, the classifier merely consider "what is the answer" without "how to get the answer", which cannot ground the answer to explicit reasoning paths. In this paper, we confront the challenge of \emph{explainable open-set} KB-VQA, where the system is required to answer questions with entities at wild and retain an explainable reasoning path. To resolve the aforementioned issues, we propose a new retriever-ranker paradigm of KB-VQA, Graph pATH rankER (GATHER for brevity). Specifically, it contains graph constructing, pruning, and path-level ranking, which not only retrieves accurate answers but also provides inference paths that explain the reasoning process. To comprehensively evaluate our model, we reformulate the benchmark dataset OK-VQA with manually corrected entity-level annotations and release it as ConceptVQA. Extensive experiments on real-world questions demonstrate that our framework is not only able to perform open-set question answering across the whole knowledge base but provide explicit reasoning path.
Authors: Ansgar Steland
Maximum-type statistics of certain functions of the sample covariance matrix of high-dimensional vector time series are studied to statistically confirm or reject the null hypothesis that a data set has been collected under normal conditions. The approach generalizes the case of the maximal deviation of the sample autocovariances function from its assumed values. Within a linear time series framework it is shown that Gumbel-type extreme value asymptotics holds true. As applications we discuss long-only mimimal-variance portfolio optimization and subportfolio analysis with respect to idiosyncratic risks, ETF index tracking by sparse tracking portfolios, convolutional deep learners for image analysis and the analysis of array-of-sensors data.
Authors: Luke Marks, Amir Abdullah, Luna Mendez, Rauno Arike, Philip Torr, Fazl Barez
Large language models (LLMs) aligned to human preferences via reinforcement learning from human feedback (RLHF) underpin many commercial applications. However, how RLHF impacts LLM internals remains opaque. We propose a novel method to interpret learned reward functions in RLHF-tuned LLMs using sparse autoencoders. Our approach trains autoencoder sets on activations from a base LLM and its RLHF-tuned version. By comparing autoencoder hidden spaces, we identify unique features that reflect the accuracy of the learned reward model. To quantify this, we construct a scenario where the tuned LLM learns token-reward mappings to maximize reward. This is the first application of sparse autoencoders for interpreting learned rewards and broadly inspecting reward learning in LLMs. Our method provides an abstract approximation of reward integrity. This presents a promising technique for ensuring alignment between specified objectives and model behaviors.
Authors: Kenan Morani
The accurate and efficient diagnosis of COVID-19 is of paramount importance, particularly in the context of large-scale medical imaging datasets. In this preprint paper, we propose a novel approach for COVID-19 diagnosis using CT images that leverages the power of Swin Transformer models, state-of-the-art solutions in computer vision tasks. Our method includes a systematic approach for patient-level predictions, where individual CT slices are classified as COVID-19 or non-COVID, and the patient's overall diagnosis is determined through majority voting. The application of the Swin Transformer in this context results in patient-level predictions that demonstrate exceptional diagnostic accuracy. In terms of evaluation metrics, our approach consistently outperforms the baseline, as well as numerous competing methods, showcasing its effectiveness in COVID-19 diagnosis. The macro F1 score achieved by our model exceeds the baseline and offers a robust solution for accurate diagnosis.
Authors: Yunus Cobanoglu
This work analyzes Graph Neural Networks, a generalization of Fully-Connected Deep Neural Nets on Graph structured data, when their width, that is the number of nodes in each fullyconnected layer is increasing to infinity. Infinite Width Neural Networks are connecting Deep Learning to Gaussian Processes and Kernels, both Machine Learning Frameworks with long traditions and extensive theoretical foundations. Gaussian Processes and Kernels have much less hyperparameters then Neural Networks and can be used for uncertainty estimation, making them more user friendly for applications. This works extends the increasing amount of research connecting Gaussian Processes and Kernels to Neural Networks. The Kernel and Gaussian Process closed forms are derived for a variety of architectures, namely the standard Graph Neural Network, the Graph Neural Network with Skip-Concatenate Connections and the Graph Attention Neural Network. All architectures are evaluated on a variety of datasets on the task of transductive Node Regression and Classification. Additionally, a Spectral Sparsification method known as Effective Resistance is used to improve runtime and memory requirements. Extending the setting to inductive graph learning tasks (Graph Regression/ Classification) is straightforward and is briefly discussed in 3.5.
Authors: Giuseppe Floris, Raffaele Mura, Luca Scionis, Giorgio Piras, Maura Pintor, Ambra Demontis, Battista Biggio
Evaluating the adversarial robustness of machine learning models using gradient-based attacks is challenging. In this work, we show that hyperparameter optimization can improve fast minimum-norm attacks by automating the selection of the loss function, the optimizer and the step-size scheduler, along with the corresponding hyperparameters. Our extensive evaluation involving several robust models demonstrates the improved efficacy of fast minimum-norm attacks when hyper-up with hyperparameter optimization. We release our open-source code at https://github.com/pralab/HO-FMN.
Authors: Qiang Li, Dan Zhang, Shengzhao Lei, Xun Zhao, Shuyan Li, Porawit Kamnoedboon, WeiWei Li
The lack of standardized robustness metrics and the widespread reliance on numerous unrelated benchmark datasets for testing have created a gap between academically validated robust models and their often problematic practical adoption. To address this, we introduce XIMAGENET-12, an explainable benchmark dataset with over 200K images and 15,600 manual semantic annotations. Covering 12 categories from ImageNet to represent objects commonly encountered in practical life and simulating six diverse scenarios, including overexposure, blurring, color changing, etc., we further propose a novel robustness criterion that extends beyond model generation ability assessment. This benchmark dataset, along with related code, is available at https://sites.google.com/view/ximagenet-12/home. Researchers and practitioners can leverage this resource to evaluate the robustness of their visual models under challenging conditions and ultimately benefit from the demands of practical computer vision systems.
Authors: Hongling Zheng, Li Shen, Anke Tang, Yong Luo, Han Hu, Bo Du, Dacheng Tao
Foundation models (FM) have demonstrated remarkable performance across a wide range of tasks (especially in the fields of natural language processing and computer vision), primarily attributed to their ability to comprehend instructions and access extensive, high-quality data. This not only showcases their current effectiveness but also sets a promising trajectory towards the development of artificial general intelligence. Unfortunately, due to multiple constraints, the raw data of the model used for large model training are often inaccessible, so the use of end-to-end models for downstream tasks has become a new research trend, which we call Learn From Model (LFM) in this article. LFM focuses on the research, modification, and design of FM based on the model interface, so as to better understand the model structure and weights (in a black box environment), and to generalize the model to downstream tasks. The study of LFM techniques can be broadly categorized into five major areas: model tuning, model distillation, model reuse, meta learning and model editing. Each category encompasses a repertoire of methods and strategies that aim to enhance the capabilities and performance of FM. This paper gives a comprehensive review of the current methods based on FM from the perspective of LFM, in order to help readers better understand the current research status and ideas. To conclude, we summarize the survey by highlighting several critical areas for future exploration and addressing open issues that require further attention from the research community. The relevant papers we investigated in this article can be accessed at <https://github.com/ruthless-man/Awesome-Learn-from-Model>.
Authors: Gokhan Budan, Francesca Damiani, Can Kurtulus, N. Kemal Ure
Model identification of battery dynamics is a central problem in energy research; many energy management systems and design processes rely on accurate battery models for efficiency optimization. The standard methodology for battery modelling is traditional design of experiments (DoE), where the battery dynamics are excited with many different current profiles and the measured outputs are used to estimate the system dynamics. However, although it is possible to obtain useful models with the traditional approach, the process is time consuming and expensive because of the need to sweep many different current-profile configurations. In the present work, a novel DoE approach is developed based on deep reinforcement learning, which alters the configuration of the experiments on the fly based on the statistics of past experiments. Instead of sticking to a library of predefined current profiles, the proposed approach modifies the current profiles dynamically by updating the output space covered by past measurements, hence only the current profiles that are informative for future experiments are applied. Simulations and real experiments are used to show that the proposed approach gives models that are as accurate as those obtained with traditional DoE but by using 85\% less resources.
Authors: Jaewoo Lee, Jaehong Yoon, Wonjae Kim, Yunji Kim, Sung Ju Hwang
We present a lifelong audio-video masked autoencoder that continually learns the multimodal representations from a video stream containing audio-video pairs, while its distribution continually shifts over time. Specifically, we propose two novel ideas to tackle the problem: (1) Localized Alignment: We introduce a small trainable multimodal encoder that predicts the audio and video tokens that are well-aligned with each other. This allows the model to learn only the highly correlated audiovisual patches with accurate multimodal relationships. (2) Forget-robust multimodal patch selection: We compare the relative importance of each audio-video patch between the current and past data pair to mitigate unintended drift of the previously learned audio-video representations. Our proposed method, FLAVA (Forget-robust Localized Audio-Video Alignment), therefore, captures the complex relationships between the audio and video modalities during training on a sequence of pre-training tasks while alleviating the forgetting of learned audiovisual correlations. Our experiments validate that FLAVA outperforms the state-of-the-art continual learning methods on several benchmark datasets under continual audio-video representation learning scenarios.
Authors: Alejandro Cholaquidis, Fabrice Gamboa, Leonardo Moreno
Regression on manifolds, and, more broadly, statistics on manifolds, has garnered significant importance in recent years due to the vast number of applications for this type of data. Circular data is a classic example, but so is data in the space of covariance matrices, data on the Grassmannian manifold obtained as a result of principal component analysis, among many others. In this work we investigate prediction sets for regression scenarios when the response variable, denoted by $Y$, resides in a manifold, and the covariable, denoted by X, lies in Euclidean space. This extends the concepts delineated in [Lei and Wasserman, 2014] to this novel context. Aligning with traditional principles in conformal inference, these prediction sets are distribution-free, indicating that no specific assumptions are imposed on the joint distribution of $(X, Y)$, and they maintain a non-parametric character. We prove the asymptotic almost sure convergence of the empirical version of these regions on the manifold to their population counterparts. The efficiency of this method is shown through a comprehensive simulation study and an analysis involving real-world data.
Authors: Bálint Mucsányi, Michael Kirchhof, Elisa Nguyen, Alexander Rubinstein, Seong Joon Oh
As machine learning technology gets applied to actual products and solutions, new challenges have emerged. Models unexpectedly fail to generalize to small changes in the distribution, tend to be confident on novel data they have never seen, or cannot communicate the rationale behind their decisions effectively with the end users. Collectively, we face a trustworthiness issue with the current machine learning technology. This textbook on Trustworthy Machine Learning (TML) covers a theoretical and technical background of four key topics in TML: Out-of-Distribution Generalization, Explainability, Uncertainty Quantification, and Evaluation of Trustworthiness. We discuss important classical and contemporary research papers of the aforementioned fields and uncover and connect their underlying intuitions. The book evolved from the homonymous course at the University of T\"ubingen, first offered in the Winter Semester of 2022/23. It is meant to be a stand-alone product accompanied by code snippets and various pointers to further sources on topics of TML. The dedicated website of the book is https://trustworthyml.io/.
Authors: Preetha Vijayan, Prashant Bhat, Elahe Arani, Bahram Zonooz
Continual learning (CL) has remained a persistent challenge for deep neural networks due to catastrophic forgetting (CF) of previously learned tasks. Several techniques such as weight regularization, experience rehearsal, and parameter isolation have been proposed to alleviate CF. Despite their relative success, these research directions have predominantly remained orthogonal and suffer from several shortcomings, while missing out on the advantages of competing strategies. On the contrary, the brain continually learns, accommodates, and transfers knowledge across tasks by simultaneously leveraging several neurophysiological processes, including neurogenesis, active forgetting, neuromodulation, metaplasticity, experience rehearsal, and context-dependent gating, rarely resulting in CF. Inspired by how the brain exploits multiple mechanisms concurrently, we propose TriRE, a novel CL paradigm that encompasses retaining the most prominent neurons for each task, revising and solidifying the extracted knowledge of current and past tasks, and actively promoting less active neurons for subsequent tasks through rewinding and relearning. Across CL settings, TriRE significantly reduces task interference and surpasses different CL approaches considered in isolation.
Authors: Minseok Choi, Chaeheon Gwak, Seho Kim, Si Hyeong Kim, Jaegul Choo
Keyphrase generation (KG) aims to generate a set of summarizing words or phrases given a source document, while keyphrase extraction (KE) aims to identify them from the text. Because the search space is much smaller in KE, it is often combined with KG to predict keyphrases that may or may not exist in the corresponding document. However, current unified approaches adopt sequence labeling and maximization-based generation that primarily operate at a token level, falling short in observing and scoring keyphrases as a whole. In this work, we propose SimCKP, a simple contrastive learning framework that consists of two stages: 1) An extractor-generator that extracts keyphrases by learning context-aware phrase-level representations in a contrastive manner while also generating keyphrases that do not appear in the document; 2) A reranker that adapts scores for each generated phrase by likewise aligning their representations with the corresponding document. Experimental results on multiple benchmark datasets demonstrate the effectiveness of our proposed approach, which outperforms the state-of-the-art models by a significant margin.
Authors: Luigi Sbailò, Luca Ghiringhelli
We observe the emergence of binary encoding within the latent space of deep-neural-network classifiers. Such binary encoding is induced by introducing a linear penultimate layer, which is equipped during training with a loss function that grows as $\exp(\vec{x}^2)$, where $\vec{x}$ are the coordinates in the latent space. The phenomenon we describe represents a specific instance of a well-documented occurrence known as \textit{neural collapse}, which arises in the terminal phase of training and entails the collapse of latent class means to the vertices of a simplex equiangular tight frame (ETF). We show that binary encoding accelerates convergence toward the simplex ETF and enhances classification accuracy.
Authors: Shaofei Cai, Bowei Zhang, Zihao Wang, Xiaojian Ma, Anji Liu, Yitao Liang
We study the problem of building a controller that can follow open-ended instructions in open-world environments. We propose to follow reference videos as instructions, which offer expressive goal specifications while eliminating the need for expensive text-gameplay annotations. A new learning framework is derived to allow learning such instruction-following controllers from gameplay videos while producing a video instruction encoder that induces a structured goal space. We implement our agent GROOT in a simple yet effective encoder-decoder architecture based on causal transformers. We evaluate GROOT against open-world counterparts and human players on a proposed Minecraft SkillForge benchmark. The Elo ratings clearly show that GROOT is closing the human-machine gap as well as exhibiting a 70% winning rate over the best generalist agent baseline. Qualitative analysis of the induced goal space further demonstrates some interesting emergent properties, including the goal composition and complex gameplay behavior synthesis. Code and video can be found on the website https://craftjarvis-groot.github.io.
Authors: Xingdong Feng, Xin He, Caixing Wang, Chao Wang, Jingnan Zhang
Covariate shift occurs prevalently in practice, where the input distributions of the source and target data are substantially different. Despite its practical importance in various learning problems, most of the existing methods only focus on some specific learning tasks and are not well validated theoretically and numerically. To tackle this problem, we propose a unified analysis of general nonparametric methods in a reproducing kernel Hilbert space (RKHS) under covariate shift. Our theoretical results are established for a general loss belonging to a rich loss function family, which includes many commonly used methods as special cases, such as mean regression, quantile regression, likelihood-based classification, and margin-based classification. Two types of covariate shift problems are the focus of this paper and the sharp convergence rates are established for a general loss function to provide a unified theoretical analysis, which concurs with the optimal results in literature where the squared loss is used. Extensive numerical studies on synthetic and real examples confirm our theoretical findings and further illustrate the effectiveness of our proposed method.
Authors: Zeyuan Ma, Hongshu Guo, Jiacheng Chen, Zhenrui Li, Guojun Peng, Yue-Jiao Gong, Yining Ma, Zhiguang Cao
Recently, Meta-Black-Box Optimization with Reinforcement Learning (MetaBBO-RL) has showcased the power of leveraging RL at the meta-level to mitigate manual fine-tuning of low-level black-box optimizers. However, this field is hindered by the lack of a unified benchmark. To fill this gap, we introduce MetaBox, the first benchmark platform expressly tailored for developing and evaluating MetaBBO-RL methods. MetaBox offers a flexible algorithmic template that allows users to effortlessly implement their unique designs within the platform. Moreover, it provides a broad spectrum of over 300 problem instances, collected from synthetic to realistic scenarios, and an extensive library of 19 baseline methods, including both traditional black-box optimizers and recent MetaBBO-RL methods. Besides, MetaBox introduces three standardized performance metrics, enabling a more thorough assessment of the methods. In a bid to illustrate the utility of MetaBox for facilitating rigorous evaluation and in-depth analysis, we carry out a wide-ranging benchmarking study on existing MetaBBO-RL methods. Our MetaBox is open-source and accessible at: https://github.com/GMC-DRL/MetaBox.
Authors: Cheongwoong Kang, Jaesik Choi
Large language models (LLMs) often make factually incorrect responses despite their success in various applications. In this paper, we hypothesize that relying heavily on simple co-occurrence statistics of the pre-training corpora is one of the main factors that cause factual errors. Our results reveal that LLMs are vulnerable to the co-occurrence bias, defined as preferring frequently co-occurred words over the correct answer. Consequently, LLMs struggle to recall facts whose subject and object rarely co-occur in the pre-training dataset although they are seen during finetuning. We show that co-occurrence bias remains despite scaling up model sizes or finetuning. Therefore, we suggest finetuning on a debiased dataset to mitigate the bias by filtering out biased samples whose subject-object co-occurrence count is high. Although debiased finetuning allows LLMs to memorize rare facts in the training set, it is not effective in recalling rare facts unseen during finetuning. Further research in mitigation will help build reliable language models by preventing potential errors. The code is available at \url{https://github.com/CheongWoong/impact_of_cooccurrence}.
Authors: Mauro Conti, Nicola Farronato, Stefanos Koffas, Luca Pajola, Stjepan Picek
Optical Character Recognition (OCR) is a widely used tool to extract text from scanned documents. Today, the state-of-the-art is achieved by exploiting deep neural networks. However, the cost of this performance is paid at the price of system vulnerability. For instance, in backdoor attacks, attackers compromise the training phase by inserting a backdoor in the victim's model that will be activated at testing time by specific patterns while leaving the overall model performance intact. This work proposes a backdoor attack for OCR resulting in the injection of non-readable characters from malicious input images. This simple but effective attack exposes the state-of-the-art OCR weakness, making the extracted text correct to human eyes but simultaneously unusable for the NLP application that uses OCR as a preprocessing step. Experimental results show that the attacked models successfully output non-readable characters for around 90% of the poisoned instances without harming their performance for the remaining instances.
Authors: Kashif Rasul, Arjun Ashok, Andrew Robert Williams, Arian Khorasani, George Adamopoulos, Rishika Bhagwatkar, Marin Biloš, Hena Ghonia, Nadhir Vincent Hassen, Anderson Schneider, Sahil Garg, Alexandre Drouin, Nicolas Chapados, Yuriy Nevmyvaka, Irina Rish
Aiming to build foundation models for time-series forecasting and study their scaling behavior, we present here our work-in-progress on Lag-Llama, a general-purpose univariate probabilistic time-series forecasting model trained on a large collection of time-series data. The model shows good zero-shot prediction capabilities on unseen "out-of-distribution" time-series datasets, outperforming supervised baselines. We use smoothly broken power-laws to fit and predict model scaling behavior. The open source code is made available at https://github.com/kashif/pytorch-transformer-ts.
Authors: Ruyi Tao, Ningning Tao, Yizhuang You, Jiang Zhang
Multiscale modeling of complex systems is crucial for understanding their intricacies. Data-driven multiscale modeling has emerged as a promising approach to tackle challenges associated with complex systems. On the other hand, self-similarity is prevalent in complex systems, hinting that large-scale complex systems can be modeled at a reduced cost. In this paper, we introduce a multiscale neural network framework that incorporates self-similarity as prior knowledge, facilitating the modeling of self-similar dynamical systems. For deterministic dynamics, our framework can discern whether the dynamics are self-similar. For uncertain dynamics, it can compare and determine which parameter set is closer to self-similarity. The framework allows us to extract scale-invariant kernels from the dynamics for modeling at any scale. Moreover, our method can identify the power law exponents in self-similar systems. Preliminary tests on the Ising model yielded critical exponents consistent with theoretical expectations, providing valuable insights for addressing critical phase transitions in non-equilibrium systems.
Authors: Olivier Laurent, Emanuel Aldea, Gianni Franchi
The distribution of the weights of modern deep neural networks (DNNs) - crucial for uncertainty quantification and robustness - is an eminently complex object due to its extremely high dimensionality. This paper proposes one of the first large-scale explorations of the posterior distribution of deep Bayesian Neural Networks (BNNs), expanding its study to real-world vision tasks and architectures. Specifically, we investigate the optimal approach for approximating the posterior, analyze the connection between posterior quality and uncertainty quantification, delve into the impact of modes on the posterior, and explore methods for visualizing the posterior. Moreover, we uncover weight-space symmetries as a critical aspect for understanding the posterior. To this extent, we develop an in-depth assessment of the impact of both permutation and scaling symmetries that tend to obfuscate the Bayesian posterior. While the first type of transformation is known for duplicating modes, we explore the relationship between the latter and L2 regularization, challenging previous misconceptions. Finally, to help the community improve our understanding of the Bayesian posterior, we will shortly release the first large-scale checkpoint dataset, including thousands of real-world models and our codes.
Authors: Arpit Mittal, Harshil Jhaveri, Swapnil Mallick, Abhishek Ajmera
Few-shot object classification is the task of classifying objects in an image with limited number of examples as supervision. We propose a one-shot/few-shot classification model that can classify an object of any unseen class into a relatively general category in an hierarchically based classification. Our model uses a three-level hierarchical contrastive loss based ResNet152 classifier for classifying an object based on its features extracted from Image embedding, not used during the training phase. For our experimentation, we have used a subset of the ImageNet (ILSVRC-12) dataset that contains only the animal classes for training our model and created our own dataset of unseen classes for evaluating our trained model. Our model provides satisfactory results in classifying the unknown objects into a generic category which has been later discussed in greater detail.
Authors: Mohamed Ashraf Abdelsalam, Samrudhdhi B. Rangrej, Isma Hadji, Nikita Dvornik, Konstantinos G. Derpanis, Afsaneh Fazly
We study the problem of future step anticipation in procedural videos. Given a video of an ongoing procedural activity, we predict a plausible next procedure step described in rich natural language. While most previous work focus on the problem of data scarcity in procedural video datasets, another core challenge of future anticipation is how to account for multiple plausible future realizations in natural settings. This problem has been largely overlooked in previous work. To address this challenge, we frame future step prediction as modelling the distribution of all possible candidates for the next step. Specifically, we design a generative model that takes a series of video clips as input, and generates multiple plausible and diverse candidates (in natural language) for the next step. Following previous work, we side-step the video annotation scarcity by pretraining our model on a large text-based corpus of procedural activities, and then transfer the model to the video domain. Our experiments, both in textual and video domains, show that our model captures diversity in the next step prediction and generates multiple plausible future predictions. Moreover, our model establishes new state-of-the-art results on YouCookII, where it outperforms existing baselines on the next step anticipation. Finally, we also show that our model can successfully transfer from text to the video domain zero-shot, ie, without fine-tuning or adaptation, and produces good-quality future step predictions from video.
Authors: Dominik Hintersdorf, Lukas Struppek, Daniel Neider, Kristian Kersting
The proliferation of large AI models trained on uncurated, often sensitive web-scraped data has raised significant privacy concerns. One of the concerns is that adversaries can extract information about the training data using privacy attacks. Unfortunately, the task of removing specific information from the models without sacrificing performance is not straightforward and has proven to be challenging. We propose a rather easy yet effective defense based on backdoor attacks to remove private information such as names of individuals from models, and focus in this work on text encoders. Specifically, through strategic insertion of backdoors, we align the embeddings of sensitive phrases with those of neutral terms-"a person" instead of the person's name. Our empirical results demonstrate the effectiveness of our backdoor-based defense on CLIP by assessing its performance using a specialized privacy attack for zero-shot classifiers. Our approach provides not only a new "dual-use" perspective on backdoor attacks, but also presents a promising avenue to enhance the privacy of individuals within models trained on uncurated web-scraped data.
Authors: Valentina Zangirolami, Matteo Borrotti
Incomplete knowledge of the environment leads an agent to make decisions under uncertainty. One of the major dilemmas in Reinforcement Learning (RL) where an autonomous agent has to balance two contrasting needs in making its decisions is: exploiting the current knowledge of the environment to maximize the cumulative reward as well as exploring actions that allow improving the knowledge of the environment, hopefully leading to higher reward values (exploration-exploitation trade-off). Concurrently, another relevant issue regards the full observability of the states, which may not be assumed in all applications. Such as when only 2D images are considered as input in a RL approach used for finding the optimal action within a 3D simulation environment. In this work, we address these issues by deploying and testing several techniques to balance exploration and exploitation trade-off on partially observable systems for predicting steering wheels in autonomous driving scenario. More precisely, the final aim is to investigate the effects of using both stochastic and deterministic multi-armed bandit strategies coupled with a Deep Recurrent Q-Network. Additionally, we adapted and evaluated the impact of an innovative method to improve the learning phase of the underlying Convolutional Recurrent Neural Network. We aim to show that adaptive stochastic methods for exploration better approximate the trade-off between exploration and exploitation as, in general, Softmax and Max-Boltzmann strategies are able to outperform epsilon-greedy techniques.
Authors: Grigory Bartosh, Dmitry Vetrov, Christian A. Naesseth
Diffusion models have shown remarkable performance on many generative tasks. Despite recent success, most diffusion models are restricted in that they only allow linear transformation of the data distribution. In contrast, broader family of transformations can potentially help train generative distributions more efficiently, simplifying the reverse process and closing the gap between the true negative log-likelihood and the variational approximation. In this paper, we present Neural Diffusion Models (NDMs), a generalization of conventional diffusion models that enables defining and learning time-dependent non-linear transformations of data. We show how to optimise NDMs using a variational bound in a simulation-free setting. Moreover, we derive a time-continuous formulation of NDMs, which allows fast and reliable inference using off-the-shelf numerical ODE and SDE solvers. Finally, we demonstrate the utility of NDMs with learnable transformations through experiments on standard image generation benchmarks, including CIFAR-10, downsampled versions of ImageNet and CelebA-HQ. NDMs outperform conventional diffusion models in terms of likelihood and produce high-quality samples.
Authors: Eve Le Guillou, Michael Will, Pierre Guillou, Jonas Lukasczyk, Pierre Fortin, Christoph Garth, Julien Tierny
This system paper presents a software framework for the support of topological analysis pipelines in a distributed-memory model. While several recent papers introduced topology-based approaches for distributed-memory environments, these were reporting experiments obtained with tailored, mono-algorithm implementations. In contrast, we describe in this paper a general-purpose, generic framework for topological analysis pipelines, i.e. a sequence of topological algorithms interacting together, possibly on distinct numbers of processes. Specifically, we instantiated our framework with the MPI model, within the Topology ToolKit (TTK). While developing this framework, we faced several algorithmic and software engineering challenges, which we document in this paper. We provide a taxonomy for the distributed-memory topological algorithms supported by TTK, depending on their communication needs and provide examples of hybrid MPI+thread parallelizations. Detailed performance analyses show that parallel efficiencies range from $20\%$ to $80\%$ (depending on the algorithms), and that the MPI-specific preconditioning introduced by our framework induces a negligible computation time overhead. We illustrate the new distributed-memory capabilities of TTK with an example of advanced analysis pipeline, combining multiple algorithms, run on the largest publicly available dataset we have found (120 billion vertices) on a standard cluster with 64 nodes (for a total of 1,536 cores). Finally, we provide a roadmap for the completion of TTK's MPI extension, along with generic recommendations for each algorithm communication category.
Authors: Yazhe Niu, Yuan Pu, Zhenjie Yang, Xueyan Li, Tong Zhou, Jiyuan Ren, Shuai Hu, Hongsheng Li, Yu Liu
Building agents based on tree-search planning capabilities with learned models has achieved remarkable success in classic decision-making problems, such as Go and Atari. However, it has been deemed challenging or even infeasible to extend Monte Carlo Tree Search (MCTS) based algorithms to diverse real-world applications, especially when these environments involve complex action spaces and significant simulation costs, or inherent stochasticity. In this work, we introduce LightZero, the first unified benchmark for deploying MCTS/MuZero in general sequential decision scenarios. Specificially, we summarize the most critical challenges in designing a general MCTS-style decision-making solver, then decompose the tightly-coupled algorithm and system design of tree-search RL methods into distinct sub-modules. By incorporating more appropriate exploration and optimization strategies, we can significantly enhance these sub-modules and construct powerful LightZero agents to tackle tasks across a wide range of domains, such as board games, Atari, MuJoCo, MiniGrid and GoBigger. Detailed benchmark results reveal the significant potential of such methods in building scalable and efficient decision intelligence. The code is available as part of OpenDILab at https://github.com/opendilab/LightZero.
Authors: Xindi Wang, Onur Varol, Tina Eliassi-Rad
Many real-world prediction tasks have outcome variables that have characteristic heavy-tail distributions. Examples include copies of books sold, auction prices of art pieces, demand for commodities in warehouses, etc. By learning heavy-tailed distributions, "big and rare" instances (e.g., the best-sellers) will have accurate predictions. Most existing approaches are not dedicated to learning heavy-tailed distribution; thus, they heavily under-predict such instances. To tackle this problem, we introduce Learning to Place (L2P), which exploits the pairwise relationships between instances for learning. In its training phase, L2P learns a pairwise preference classifier: is instance A > instance B? In its placing phase, L2P obtains a prediction by placing the new instance among the known instances. Based on its placement, the new instance is then assigned a value for its outcome variable. Experiments on real data show that L2P outperforms competing approaches in terms of accuracy and ability to reproduce heavy-tailed outcome distribution. In addition, L2P provides an interpretable model by placing each predicted instance in relation to its comparable neighbors. Interpretable models are highly desirable when lives and treasure are at stake.
Authors: Yijia Wang, Matthias Poloczek, Daniel R. Jiang
Reinforcement learning in sparse-reward navigation environments with expensive and limited interactions is challenging and poses a need for effective exploration. Motivated by complex navigation tasks that require real-world training (when cheap simulators are not available), we consider an agent that faces an unknown distribution of environments and must decide on an exploration strategy. It may leverage a series of training environments to improve its policy before it is evaluated in a test environment drawn from the same environment distribution. Most existing approaches focus on fixed exploration strategies, while the few that view exploration as a meta-optimization problem tend to ignore the need for cost-efficient exploration. We propose a cost-aware Bayesian optimization approach that efficiently searches over a class of dynamic subgoal-based exploration strategies. The algorithm adjusts a variety of levers -- the locations of the subgoals, the length of each episode, and the number of replications per trial -- in order to overcome the challenges of sparse rewards, expensive interactions, and noise. An experimental evaluation demonstrates that the new approach outperforms existing baselines across a number of problem domains. We also provide a theoretical foundation and prove that the method asymptotically identifies a near-optimal subgoal design.
Authors: Shujian Liao, Hao Ni, Lukasz Szpruch, Magnus Wiese, Marc Sabate-Vidales, Baoren Xiao
Generative adversarial networks (GANs) have been extremely successful in generating samples, from seemingly high dimensional probability measures. However, these methods struggle to capture the temporal dependence of joint probability distributions induced by time-series data. Furthermore, long time-series data streams hugely increase the dimension of the target space, which may render generative modelling infeasible. To overcome these challenges, motivated by the autoregressive models in econometric, we are interested in the conditional distribution of future time series given the past information. We propose the generic conditional Sig-WGAN framework by integrating Wasserstein-GANs (WGANs) with mathematically principled and efficient path feature extraction called the signature of a path. The signature of a path is a graded sequence of statistics that provides a universal description for a stream of data, and its expected value characterises the law of the time-series model. In particular, we develop the conditional Sig-$W_1$ metric, that captures the conditional joint law of time series models, and use it as a discriminator. The signature feature space enables the explicit representation of the proposed discriminators which alleviates the need for expensive training. We validate our method on both synthetic and empirical dataset and observe that our method consistently and significantly outperforms state-of-the-art benchmarks with respect to measures of similarity and predictive ability.
Authors: Yun Yue, Yongchao Liu, Suo Tong, Minghao Li, Zhen Zhang, Chunyang Wen, Huanjun Bao, Lihong Gu, Jinjie Gu, Yixiang Mu
We develop a novel framework that adds the regularizers of the sparse group lasso to a family of adaptive optimizers in deep learning, such as Momentum, Adagrad, Adam, AMSGrad, AdaHessian, and create a new class of optimizers, which are named Group Momentum, Group Adagrad, Group Adam, Group AMSGrad and Group AdaHessian, etc., accordingly. We establish theoretically proven convergence guarantees in the stochastic convex settings, based on primal-dual methods. We evaluate the regularized effect of our new optimizers on three large-scale real-world ad click datasets with state-of-the-art deep learning models. The experimental results reveal that compared with the original optimizers with the post-processing procedure which uses the magnitude pruning method, the performance of the models can be significantly improved on the same sparsity level. Furthermore, in comparison to the cases without magnitude pruning, our methods can achieve extremely high sparsity with significantly better or highly competitive performance. The code is available at https://github.com/intelligent-machine-learning/dlrover/blob/master/tfplus.
Authors: Michael Kamp, Jonas Fischer, Jilles Vreeken
Federated learning allows multiple parties to collaboratively train a joint model without sharing local data. This enables applications of machine learning in settings of inherently distributed, undisclosable data such as in the medical domain. In practice, joint training is usually achieved by aggregating local models, for which local training objectives have to be in expectation similar to the joint (global) objective. Often, however, local datasets are so small that local objectives differ greatly from the global objective, resulting in federated learning to fail. We propose a novel approach that intertwines model aggregations with permutations of local models. The permutations expose each local model to a daisy chain of local datasets resulting in more efficient training in data-sparse domains. This enables training on extremely small local datasets, such as patient data across hospitals, while retaining the training efficiency and privacy benefits of federated learning.
Authors: Tarun Kalluri, Astuti Sharma, Manmohan Chandraker
Practical real world datasets with plentiful categories introduce new challenges for unsupervised domain adaptation like small inter-class discriminability, that existing approaches relying on domain invariance alone cannot handle sufficiently well. In this work we propose MemSAC, which exploits sample level similarity across source and target domains to achieve discriminative transfer, along with architectures that scale to a large number of categories. For this purpose, we first introduce a memory augmented approach to efficiently extract pairwise similarity relations between labeled source and unlabeled target domain instances, suited to handle an arbitrary number of classes. Next, we propose and theoretically justify a novel variant of the contrastive loss to promote local consistency among within-class cross domain samples while enforcing separation between classes, thus preserving discriminative transfer from source to target. We validate the advantages of MemSAC with significant improvements over previous state-of-the-art on multiple challenging transfer tasks designed for large-scale adaptation, such as DomainNet with 345 classes and fine-grained adaptation on Caltech-UCSD birds dataset with 200 classes. We also provide in-depth analysis and insights into the effectiveness of MemSAC.
Authors: Martim Sousa, Ana Maria Tomé, José Moreira
This paper introduces a novel model-agnostic algorithm called adaptive ensemble batch multi-input multi-output conformalized quantile regression (AEnbMIMOCQR} that enables forecasters to generate multi-step ahead prediction intervals for a fixed pre-specified miscoverage rate in a distribution-free manner. Our method is grounded on conformal prediction principles, however, it does not require data splitting and provides close to exact coverage even when the data is not exchangeable. Moreover, the resulting prediction intervals, besides being empirically valid along the forecast horizon, do not neglect heteroscedasticity. AEnbMIMOCQR is designed to be robust to distribution shifts, which means that its prediction intervals remain reliable over an unlimited period of time, without entailing retraining or imposing unrealistic strict assumptions on the data-generating process. Through methodically experimentation, we demonstrate that our approach outperforms other competitive methods on both real-world and synthetic datasets. The code used in the experimental part and a tutorial on how to use AEnbMIMOCQR can be found at the following GitHub repository: https://github.com/Quilograma/AEnbMIMOCQR.
Authors: Rob Brekelmans, Frank Nielsen
Markov Chain Monte Carlo methods for sampling from complex distributions and estimating normalization constants often simulate samples from a sequence of intermediate distributions along an annealing path, which bridges between a tractable initial distribution and a target density of interest. Prior work has constructed annealing paths using quasi-arithmetic means, and interpreted the resulting intermediate densities as minimizing an expected divergence to the endpoints. We provide a comprehensive analysis of this 'centroid' property using Bregman divergences under a monotonic embedding of the density function, thereby associating common divergences such as Amari's and Renyi's ${\alpha}$-divergences, ${(\alpha,\beta)}$-divergences, and the Jensen-Shannon divergence with intermediate densities along an annealing path. Our analysis highlights the interplay between parametric families, quasi-arithmetic means, and divergence functions using the rho-tau Bregman divergence framework of Zhang 2004,2013.
Authors: Anna Konrad, John McDonald, Rudi Villing
We present the Grasp Proposal Network (GP-net), a Convolutional Neural Network model which can generate 6-DoF grasps from flexible viewpoints, e.g. as experienced by mobile manipulators. To train GP-net, we synthetically generate a dataset containing depth-images and ground-truth grasp information. In real-world experiments, we use the EGAD evaluation benchmark to evaluate GP-net against two commonly used algorithms, the Volumetric Grasping Network (VGN) and the Grasp Pose Detection package (GPD), on a PAL TIAGo mobile manipulator. In contrast to the state-of-the-art methods in robotic grasping, GP-net can be used for grasping objects from flexible, unknown viewpoints without the need to define the workspace and achieves a grasp success of 54.4% compared to 51.6% for VGN and 44.2% for GPD. We provide a ROS package along with our code and pre-trained models at https://aucoroboticsmu.github.io/GP-net/.
Authors: Lorenzo Zambon, Dario Azzimonti, Giorgio Corani
Hierarchical time series are common in several applied fields. The forecasts for these time series are required to be coherent, that is, to satisfy the constraints given by the hierarchy. The most popular technique to enforce coherence is called reconciliation, which adjusts the base forecasts computed for each time series. However, recent works on probabilistic reconciliation present several limitations. In this paper, we propose a new approach based on conditioning to reconcile any type of forecast distribution. We then introduce a new algorithm, called Bottom-Up Importance Sampling, to efficiently sample from the reconciled distribution. It can be used for any base forecast distribution: discrete, continuous, or in the form of samples, providing a major speedup compared to the current methods. Experiments on several temporal hierarchies show a significant improvement over base probabilistic forecasts.
Authors: Ziqi Gao, Yuntao Wang, Jianguo Chen, Junliang Xing, Shwetak Patel, Xin Liu, Yuanchun Shi
Multimodal sensors provide complementary information to develop accurate machine-learning methods for human activity recognition (HAR), but introduce significantly higher computational load, which reduces efficiency. This paper proposes an efficient multimodal neural architecture for HAR using an RGB camera and inertial measurement units (IMUs) called Multimodal Temporal Segment Attention Network (MMTSA). MMTSA first transforms IMU sensor data into a temporal and structure-preserving gray-scale image using the Gramian Angular Field (GAF), representing the inherent properties of human activities. MMTSA then applies a multimodal sparse sampling method to reduce data redundancy. Lastly, MMTSA adopts an inter-segment attention module for efficient multimodal fusion. Using three well-established public datasets, we evaluated MMTSA's effectiveness and efficiency in HAR. Results show that our method achieves superior performance improvements 11.13% of cross-subject F1-score on the MMAct dataset than the previous state-of-the-art (SOTA) methods. The ablation study and analysis suggest that MMTSA's effectiveness in fusing multimodal data for accurate HAR. The efficiency evaluation on an edge device showed that MMTSA achieved significantly better accuracy, lower computational load, and lower inference latency than SOTA methods.
Authors: Anish Agarwal, Sarah H. Cen, Devavrat Shah, Christina Lee Yu
We propose a generalization of the synthetic controls and synthetic interventions methodology to incorporate network interference. We consider the estimation of unit-specific potential outcomes from panel data in the presence of spillover across units and unobserved confounding. Key to our approach is a novel latent factor model that takes into account network interference and generalizes the factor models typically used in panel data settings. We propose an estimator, Network Synthetic Interventions (NSI), and show that it consistently estimates the mean outcomes for a unit under an arbitrary set of counterfactual treatments for the network. We further establish that the estimator is asymptotically normal. We furnish two validity tests for whether the NSI estimator reliably generalizes to produce accurate counterfactual estimates. We provide a novel graph-based experiment design that guarantees the NSI estimator produces accurate counterfactual estimates, and also analyze the sample complexity of the proposed design. We conclude with simulations that corroborate our theoretical findings.
Authors: Ying Yuan, Giovanni Apruzzese, Mauro Conti
Existing literature on adversarial Machine Learning (ML) focuses either on showing attacks that break every ML model, or defenses that withstand most attacks. Unfortunately, little consideration is given to the actual feasibility of the attack or the defense. Moreover, adversarial samples are often crafted in the "feature-space", making the corresponding evaluations of questionable value. Simply put, the current situation does not allow to estimate the actual threat posed by adversarial attacks, leading to a lack of secure ML systems.
We aim to clarify such confusion in this paper. By considering the application of ML for Phishing Website Detection (PWD), we formalize the "evasion-space" in which an adversarial perturbation can be introduced to fool a ML-PWD -- demonstrating that even perturbations in the "feature-space" are useful. Then, we propose a realistic threat model describing evasion attacks against ML-PWD that are cheap to stage, and hence intrinsically more attractive for real phishers. After that, we perform the first statistically validated assessment of state-of-the-art ML-PWD against 12 evasion attacks. Our evaluation shows (i) the true efficacy of evasion attempts that are more likely to occur; and (ii) the impact of perturbations crafted in different evasion-spaces. Our realistic evasion attempts induce a statistically significant degradation (3-10% at p<0.05), and their cheap cost makes them a subtle threat. Notably, however, some ML-PWD are immune to our most realistic attacks (p=0.22).
Finally, as an additional contribution of this journal publication, we are the first to consider the intriguing case wherein an attacker introduces perturbations in multiple evasion-spaces at the same time. These new results show that simultaneously applying perturbations in the problem- and feature-space can cause a drop in the detection rate from 0.95 to 0.
Authors: Wenguan Wang, Yi Yang, Fei Wu
Neural-symbolic computing (NeSy), which pursues the integration of the symbolic and statistical paradigms of cognition, has been an active research area of Artificial Intelligence (AI) for many years. As NeSy shows promise of reconciling the advantages of reasoning and interpretability of symbolic representation and robust learning in neural networks, it may serve as a catalyst for the next generation of AI. In the present paper, we provide a systematic overview of the recent developments and important contributions of NeSy research. Firstly, we introduce study history of this area, covering early work and foundations. We further discuss background concepts and identify key driving factors behind the development of NeSy. Afterward, we categorize recent landmark approaches along several main characteristics that underline this research paradigm, including neural-symbolic integration, knowledge representation, knowledge embedding, and functionality. Next, we briefly discuss the successful application of modern NeSy approaches in several domains. Then, we benchmark several NeSy methods on three representative application tasks. Finally, we identify the open problems together with potential future research directions. This survey is expected to help new researchers enter this rapidly evolving field and accelerate the progress towards data-and knowledge-driven AI.
Authors: Adrian Goldwaser, Hong Ge
Larger and deeper networks generalise well despite their increased capacity to overfit. Understanding why this happens is theoretically and practically important. One recent approach looks at the infinitely wide limits of such networks and their corresponding kernels. However, these theoretical tools cannot fully explain finite networks as the empirical kernel changes significantly during gradient-descent-based training in contrast to infinite networks. In this work, we derive an iterative linearised training method as a novel empirical tool to further investigate this distinction, allowing us to control for sparse (i.e. infrequent) feature updates and quantify the frequency of feature learning needed to achieve comparable performance. We justify iterative linearisation as an interpolation between a finite analog of the infinite width regime, which does not learn features, and standard gradient descent training, which does. Informally, we also show that it is analogous to a damped version of the Gauss-Newton algorithm -- a second-order method. We show that in a variety of cases, iterative linearised training surprisingly performs on par with standard training, noting in particular how much less frequent feature learning is required to achieve comparable performance. We also show that feature learning is essential for good performance. Since such feature learning inevitably causes changes in the NTK kernel, we provide direct negative evidence for the NTK theory, which states the NTK kernel remains constant during training.
Authors: Xutan Peng, Yipeng Zhang, Jingfeng Yang, Mark Stevenson
Although it has been demonstrated that Natural Language Processing (NLP) algorithms are vulnerable to deliberate attacks, the question of whether such weaknesses can lead to software security threats is under-explored. To bridge this gap, we conducted vulnerability tests on Text-to-SQL systems that are commonly used to create natural language interfaces to databases. We showed that the Text-to-SQL modules within six commercial applications can be manipulated to produce malicious code, potentially leading to data breaches and Denial of Service attacks. This is the first demonstration that NLP models can be exploited as attack vectors in the wild. In addition, experiments using four open-source language models verified that straightforward backdoor attacks on Text-to-SQL systems achieve a 100% success rate without affecting their performance. The aim of this work is to draw the community's attention to potential software security issues associated with NLP algorithms and encourage exploration of methods to mitigate against them.
Authors: Alison Pouplin, David Eklund, Carl Henrik Ek, Søren Hauberg
Riemannian geometry provides us with powerful tools to explore the latent space of generative models while preserving the underlying structure of the data. The latent space can be equipped it with a Riemannian metric, pulled back from the data manifold. With this metric, we can systematically navigate the space relying on geodesics defined as the shortest curves between two points. Generative models are often stochastic, causing the data space, the Riemannian metric, and the geodesics, to be stochastic as well. Stochastic objects are at best impractical, and at worst impossible, to manipulate. A common solution is to approximate the stochastic pullback metric by its expectation. But the geodesics derived from this expected Riemannian metric do not correspond to the expected length-minimising curves. In this work, we propose another metric whose geodesics explicitly minimise the expected length of the pullback metric. We show this metric defines a Finsler metric, and we compare it with the expected Riemannian metric. In high dimensions, we prove that both metrics converge to each other at a rate of $O\left(\frac{1}{D}\right)$. This convergence implies that the established expected Riemannian metric is an accurate approximation of the theoretically more grounded Finsler metric. This provides justification for using the expected Riemannian metric for practical implementations.
Authors: Mingqi Yuan, Bo Li, Xin Jin, Wenjun Zeng
We present AIRS: Automatic Intrinsic Reward Shaping that intelligently and adaptively provides high-quality intrinsic rewards to enhance exploration in reinforcement learning (RL). More specifically, AIRS selects shaping function from a predefined set based on the estimated task return in real-time, providing reliable exploration incentives and alleviating the biased objective problem. Moreover, we develop an intrinsic reward toolkit to provide efficient and reliable implementations of diverse intrinsic reward approaches. We test AIRS on various tasks of MiniGrid, Procgen, and DeepMind Control Suite. Extensive simulation demonstrates that AIRS can outperform the benchmarking schemes and achieve superior performance with simple architecture.
Authors: Zhanglu Yan, Shida Wang, Kaiwen Tang, Weng-Fai Wong
Hyperdimensional computing (HDC) is a method to perform classification that uses binary vectors with high dimensions and the majority rule. This approach has the potential to be energy-efficient and hence deemed suitable for resource-limited platforms due to its simplicity and massive parallelism. However, in order to achieve high accuracy, HDC sometimes uses hypervectors with tens of thousands of dimensions. This potentially negates its efficiency advantage. In this paper, we examine the necessity of such high dimensions and conduct a detailed theoretical analysis of the relationship between hypervector dimensions and accuracy. Our results demonstrate that as the dimension of the hypervectors increases, the worst-case/average-case HDC prediction accuracy with the majority rule decreases. Building on this insight, we develop HDC models that use binary hypervectors with dimensions orders of magnitude lower than those of state-of-the-art HDC models while maintaining equivalent or even improved accuracy and efficiency. For instance, on the MNIST dataset, we achieve 91.12% HDC accuracy in image classification with a dimension of only 64. Our methods perform operations that are only 0.35% of other HDC models with dimensions of 10,000. Furthermore, we evaluate our methods on ISOLET, UCI-HAR, and Fashion-MNIST datasets and investigate the limits of HDC computing.
Authors: Guanyu Nie, Yididiya Y Nadew, Yanhui Zhu, Vaneet Aggarwal, Christopher John Quinn
We investigate the problem of stochastic, combinatorial multi-armed bandits where the learner only has access to bandit feedback and the reward function can be non-linear. We provide a general framework for adapting discrete offline approximation algorithms into sublinear $\alpha$-regret methods that only require bandit feedback, achieving $\mathcal{O}\left(T^\frac{2}{3}\log(T)^\frac{1}{3}\right)$ expected cumulative $\alpha$-regret dependence on the horizon $T$. The framework only requires the offline algorithms to be robust to small errors in function evaluation. The adaptation procedure does not even require explicit knowledge of the offline approximation algorithm -- the offline algorithm can be used as a black box subroutine. To demonstrate the utility of the proposed framework, the proposed framework is applied to diverse applications in submodular maximization. The new CMAB algorithms for submodular maximization with knapsack constraints outperform a full-bandit method developed for the adversarial setting in experiments with real-world data.
Authors: Anton Orlichenko, Grant Daly, Ziyu Zhou, Anqi Liu, Hui Shen, Hong-Wen Deng, Yu-Ping Wang
Most packages for the analysis of fMRI-based functional connectivity (FC) and genomic data are used with a programming language interface, lacking an easy-to-navigate GUI frontend. This exacerbates two problems found in these types of data: demographic confounds and quality control in the face of high dimensionality of features. The reason is that it is too slow and cumbersome to use a programming interface to create all the necessary visualizations required to identify all correlations, confounding effects, or quality control problems in a dataset. To remedy this situation, we have developed ImageNomer, a data visualization and analysis tool that allows inspection of both subject-level and cohort-level demographic, genomic, and imaging features. The software is Python-based, runs in a self-contained Docker image, and contains a browser-based GUI frontend. We demonstrate the usefulness of ImageNomer by identifying an unexpected race confound when predicting achievement scores in the Philadelphia Neurodevelopmental Cohort (PNC) dataset. In the past, many studies have attempted to use FC to identify achievement-related features in fMRI. Using ImageNomer, we find a clear potential for confounding effects of race. Using correlation analysis in the ImageNomer software, we show that FCs correlated with Wide Range Achievement Test (WRAT) score are in fact more highly correlated with race. Investigating further, we find that whereas both FC and SNP (genomic) features can account for 10-15\% of WRAT score variation, this predictive ability disappears when controlling for race. In this work, we demonstrate the advantage of our ImageNomer GUI tool in data exploration and confound detection. Additionally, this work identifies race as a strong confound in FC data and casts doubt on the possibility of finding unbiased achievement-related features in fMRI and SNP data of healthy adolescents.
Authors: Amrith Setlur, Don Dennis, Benjamin Eysenbach, Aditi Raghunathan, Chelsea Finn, Virginia Smith, Sergey Levine
Training machine learning models robust to distribution shifts is critical for real-world applications. Some robust training algorithms (e.g., Group DRO) specialize to group shifts and require group information on all training points. Other methods (e.g., CVaR DRO) that do not need group annotations can be overly conservative, since they naively upweight high loss points which may form a contrived set that does not correspond to any meaningful group in the real world (e.g., when the high loss points are randomly mislabeled training points). In this work, we address limitations in prior approaches by assuming a more nuanced form of group shift: conditioned on the label, we assume that the true group function (indicator over group) is simple. For example, we may expect that group shifts occur along low bitrate features (e.g., image background, lighting). Thus, we aim to learn a model that maintains high accuracy on simple group functions realized by these low bitrate features, that need not spend valuable model capacity achieving high accuracy on contrived groups of examples. Based on this, we consider the two-player game formulation of DRO where the adversary's capacity is bitrate-constrained. Our resulting practical algorithm, Bitrate-Constrained DRO (BR-DRO), does not require group information on training samples yet matches the performance of Group DRO on datasets that have training group annotations and that of CVaR DRO on long-tailed distributions. Our theoretical analysis reveals that in some settings BR-DRO objective can provably yield statistically efficient and less conservative solutions than unconstrained CVaR DRO.
Authors: Hanlin Zhu, Amy Zhang
Goal-conditioned reinforcement learning (GCRL) refers to learning general-purpose skills that aim to reach diverse goals. In particular, offline GCRL only requires purely pre-collected datasets to perform training tasks without additional interactions with the environment. Although offline GCRL has become increasingly prevalent and many previous works have demonstrated its empirical success, the theoretical understanding of efficient offline GCRL algorithms is not well established, especially when the state space is huge and the offline dataset only covers the policy we aim to learn. In this paper, we provide a rigorous theoretical analysis of an existing empirically successful offline GCRL algorithm. We prove that under slight modification, this algorithm enjoys an $\widetilde{O}(\text{poly}(1/\epsilon))$ sample complexity (where $\epsilon$ is the desired suboptimality of the learned policy) with general function approximation thanks to the property of (semi-)strong convexity of the objective functions. We only require nearly minimal assumptions on the dataset (single-policy concentrability) and the function class (realizability). Moreover, this algorithm consists of two uninterleaved optimization steps, which we refer to as $V$-learning and policy learning, and is computationally stable since it does not involve minimax optimization. We also empirically validate our theory by showing that the modified algorithm outperforms the previous algorithm in various real-world environments. To the best of our knowledge, this is the first algorithm that is both provably efficient with general function approximation and single-policy concentrability, and empirically successful without requiring solving minimax optimization problems.
Authors: Hailey Joren, Chirag Nagpal, Katherine Heller, Berk Ustun
Machine learning models are often personalized with information that is protected, sensitive, self-reported, or costly to acquire. These models use information about people but do not facilitate nor inform their consent. Individuals cannot opt out of reporting personal information to a model, nor tell if they benefit from personalization in the first place. We introduce a family of classification models, called participatory systems, that let individuals opt into personalization at prediction time. We present a model-agnostic algorithm to learn participatory systems for personalization with categorical group attributes. We conduct a comprehensive empirical study of participatory systems in clinical prediction tasks, benchmarking them with common approaches for personalization and imputation. Our results demonstrate that participatory systems can facilitate and inform consent while improving performance and data use across all groups who report personal data.
Authors: Max Lamparth, Anka Reuel
Poisoning of data sets is a potential security threat to large language models that can lead to backdoored models. A description of the internal mechanisms of backdoored language models and how they process trigger inputs, e.g., when switching to toxic language, has yet to be found. In this work, we study the internal representations of transformer-based backdoored language models and determine early-layer MLP modules as most important for the backdoor mechanism in combination with the initial embedding projection. We use this knowledge to remove, insert, and modify backdoor mechanisms with engineered replacements that reduce the MLP module outputs to essentials for the backdoor mechanism. To this end, we introduce PCP ablation, where we replace transformer modules with low-rank matrices based on the principal components of their activations. We demonstrate our results on backdoored toy, backdoored large, and non-backdoored open-source models. We show that we can improve the backdoor robustness of large language models by locally constraining individual modules during fine-tuning on potentially poisonous data sets.
Trigger warning: Offensive language.
Authors: Kushagra Pandey, Stephan Mandt
Score-based Generative Models (SGMs) have demonstrated exceptional synthesis outcomes across various tasks. However, the current design landscape of the forward diffusion process remains largely untapped and often relies on physical heuristics or simplifying assumptions. Utilizing insights from the development of scalable Bayesian posterior samplers, we present a complete recipe for formulating forward processes in SGMs, ensuring convergence to the desired target distribution. Our approach reveals that several existing SGMs can be seen as specific manifestations of our framework. Building upon this method, we introduce Phase Space Langevin Diffusion (PSLD), which relies on score-based modeling within an augmented space enriched by auxiliary variables akin to physical phase space. Empirical results exhibit the superior sample quality and improved speed-quality trade-off of PSLD compared to various competing approaches on established image synthesis benchmarks. Remarkably, PSLD achieves sample quality akin to state-of-the-art SGMs (FID: 2.10 for unconditional CIFAR-10 generation). Lastly, we demonstrate the applicability of PSLD in conditional synthesis using pre-trained score networks, offering an appealing alternative as an SGM backbone for future advancements. Code and model checkpoints can be accessed at \url{https://github.com/mandt-lab/PSLD}.
Authors: Tina Dorosti, Manuel Schultheiss, Felix Hofmann, Johannes Thalhammer, Luisa Kirchner, Theresa Urban, Franz Pfeiffer, Florian Schaff, Tobias Lasser, Daniela Pfeiffer
We aim to optimize the binary detection of Chronic Obstructive Pulmonary Disease (COPD) based on emphysema presence in the lung with convolutional neural networks (CNN) by exploring manually adjusted versus automated window-setting optimization (WSO) on computed tomography (CT) images. 7,194 CT images (3,597 with COPD; 3,597 healthy controls) from 78 subjects (43 with COPD; 35 healthy controls) were selected retrospectively (10.2018-12.2019) and preprocessed. For each image, intensity values were manually clipped to the emphysema window setting and a baseline 'full-range' window setting. Class-balanced train, validation, and test sets contained 3,392, 1,114, and 2,688 images. The network backbone was optimized by comparing various CNN architectures. Furthermore, automated WSO was implemented by adding a customized layer to the model. The image-level area under the Receiver Operating Characteristics curve (AUC) [lower, upper limit 95% confidence] was utilized to compare model variations. Repeated inference (n=7) on the test set showed that the DenseNet was the most efficient backbone and achieved a mean AUC of 0.80 [0.76, 0.85] without WSO. Comparably, with input images manually adjusted to the emphysema window, the DenseNet model predicted COPD with a mean AUC of 0.86 [0.82, 0.89]. By adding a customized WSO layer to the DenseNet, an optimal window in the proximity of the emphysema window setting was learned automatically, and a mean AUC of 0.82 [0.78, 0.86] was achieved. Detection of COPD with DenseNet models was improved by WSO of CT data to the emphysema window setting range.
Authors: Aadirupa Saha, Branislav Kveton
Most bandit algorithms assume that the reward variances or their upper bounds are known, and that they are the same for all arms. This naturally leads to suboptimal performance and higher regret due to variance overestimation. On the other hand, underestimated reward variances may lead to linear regret due to committing early to a suboptimal arm. This motivated prior works on variance-adaptive frequentist algorithms, which have strong instance-dependent regret bounds but cannot incorporate prior knowledge on reward variances. We lay foundations for the Bayesian setting, which incorporates prior knowledge. This results in lower regret in practice, due to using the prior in the algorithm design, and also improved regret guarantees. Specifically, we study Gaussian bandits with {unknown heterogeneous reward variances}, and develop a Thompson sampling algorithm with prior-dependent Bayes regret bounds. We achieve lower regret with lower reward variances and more informative priors on them, which is precisely why we pay only for what is uncertain. This is the first result of its kind. Finally, we corroborate our theory with extensive experiments, which show the superiority of our variance-adaptive Bayesian algorithm over prior frequentist approaches. We also show that our approach is robust to model misspecification and can be applied with estimated priors.
Authors: Gang Liu, Eric Inae, Tong Zhao, Jiaxin Xu, Tengfei Luo, Meng Jiang
Graph property prediction tasks are important and numerous. While each task offers a small size of labeled examples, unlabeled graphs have been collected from various sources and at a large scale. A conventional approach is training a model with the unlabeled graphs on self-supervised tasks and then fine-tuning the model on the prediction tasks. However, the self-supervised task knowledge could not be aligned or sometimes conflicted with what the predictions needed. In this paper, we propose to extract the knowledge underlying the large set of unlabeled graphs as a specific set of useful data points to augment each property prediction model. We use a diffusion model to fully utilize the unlabeled graphs and design two new objectives to guide the model's denoising process with each task's labeled data to generate task-specific graph examples and their labels. Experiments demonstrate that our data-centric approach performs significantly better than fifteen existing various methods on fifteen tasks. The performance improvement brought by unlabeled data is visible as the generated labeled examples unlike the self-supervised learning.
Authors: Akifumi Okuno, Kazuharu Harada
This study proposes an interpretable neural network-based non-proportional odds model (N$^3$POM) for ordinal regression. N$^3$POM is different from conventional approaches to ordinal regression with non-proportional models in several ways: (1) N$^3$POM is designed to directly handle continuous responses, whereas standard methods typically treat de facto ordered continuous variables as discrete, (2) instead of estimating response-dependent finite coefficients of linear models from discrete responses as is done in conventional approaches, we train a non-linear neural network to serve as a coefficient function. Thanks to the neural network, N$^3$POM offers flexibility while preserving the interpretability of conventional ordinal regression. We establish a sufficient condition under which the predicted conditional cumulative probability locally satisfies the monotonicity constraint over a user-specified region in the covariate space. Additionally, we provide a monotonicity-preserving stochastic (MPS) algorithm for effectively training the neural network. We apply N$^3$POM to several real-world datasets.
Authors: Patrik Puchert, Poonam Poonam, Christian van Onzenoodt, Timo Ropinski
Large Language Models (LLMs) have revolutionized natural language processing and demonstrated impressive capabilities in various tasks. Unfortunately, they are prone to hallucinations, where the model exposes incorrect or false information in its responses, which renders diligent evaluation approaches mandatory. While LLM performance in specific knowledge fields is often evaluated based on question and answer (Q&A) datasets, such evaluations usually report only a single accuracy number for the dataset, which often covers an entire field. This field-based evaluation, is problematic with respect to transparency and model improvement. A stratified evaluation could instead reveal subfields, where hallucinations are more likely to occur and thus help to better assess LLMs' risks and guide their further development. To support such stratified evaluations, we propose LLMMaps as a novel visualization technique that enables users to evaluate LLMs' performance with respect to Q&A datasets. LLMMaps provide detailed insights into LLMs' knowledge capabilities in different subfields, by transforming Q&A datasets as well as LLM responses into an internal knowledge structure. An extension for comparative visualization furthermore, allows for the detailed comparison of multiple LLMs. To assess LLMMaps we use them to conduct a comparative analysis of several state-of-the-art LLMs, such as BLOOM, GPT-2, GPT-3, ChatGPT and LLaMa-13B, as well as two qualitative user evaluations. All necessary source code and data for generating LLMMaps to be used in scientific publications and elsewhere is available on GitHub: https://github.com/viscom-ulm/LLMMaps
Authors: Arvi Jonnarth, Yushan Zhang, Michael Felsberg
Image-level weakly-supervised semantic segmentation (WSSS) reduces the usually vast data annotation cost by surrogate segmentation masks during training. The typical approach involves training an image classification network using global average pooling (GAP) on convolutional feature maps. This enables the estimation of object locations based on class activation maps (CAMs), which identify the importance of image regions. The CAMs are then used to generate pseudo-labels, in the form of segmentation masks, to supervise a segmentation model in the absence of pixel-level ground truth. Our work is based on two techniques for improving CAMs; importance sampling, which is a substitute for GAP, and the feature similarity loss, which utilizes a heuristic that object contours almost always align with color edges in images. However, both are based on the multinomial posterior with softmax, and implicitly assume that classes are mutually exclusive, which turns out suboptimal in our experiments. Thus, we reformulate both techniques based on binomial posteriors of multiple independent binary problems. This has two benefits; their performance is improved and they become more general, resulting in an add-on method that can boost virtually any WSSS method. This is demonstrated on a wide variety of baselines on the PASCAL VOC dataset, improving the region similarity and contour quality of all implemented state-of-the-art methods. Experiments on the MS COCO dataset show that our proposed add-on is well-suited for large-scale settings. Our code is available at https://github.com/arvijj/hfpl.
Authors: Tengfei Xu, Dachuan Liu, Peng Hao, Bo Wang
Neural operators as novel neural architectures for fast approximating solution operators of partial differential equations (PDEs), have shown considerable promise for future scientific computing. However, the mainstream of training neural operators is still data-driven, which needs an expensive ground-truth dataset from various sources (e.g., solving PDEs' samples with the conventional solvers, real-world experiments) in addition to training stage costs. From a computational perspective, marrying operator learning and specific domain knowledge to solve PDEs is an essential step in reducing dataset costs and label-free learning. We propose a novel paradigm that provides a unified framework of training neural operators and solving PDEs with the variational form, which we refer to as the variational operator learning (VOL). Ritz and Galerkin approach with finite element discretization are developed for VOL to achieve matrix-free approximation of system functional and residual, then direct minimization and iterative update are proposed as two optimization strategies for VOL. Various types of experiments based on reasonable benchmarks about variable heat source, Darcy flow, and variable stiffness elasticity are conducted to demonstrate the effectiveness of VOL. With a label-free training set and a 5-label-only shift set, VOL learns solution operators with its test errors decreasing in a power law with respect to the amount of unlabeled data. To the best of the authors' knowledge, this is the first study that integrates the perspectives of the weak form and efficient iterative methods for solving sparse linear systems into the end-to-end operator learning task.
Authors: Jonah Botvinick-Greenhouse, Yunan Yang, Romit Maulik
Motivated by the computational difficulties incurred by popular deep learning algorithms for the generative modeling of temporal densities, we propose a cheap alternative which requires minimal hyperparameter tuning and scales favorably to high dimensional problems. In particular, we use a projection-based optimal transport solver [Meng et al., 2019] to join successive samples and subsequently use transport splines [Chewi et al., 2020] to interpolate the evolving density. When the sampling frequency is sufficiently high, the optimal maps are close to the identity and are thus computationally efficient to compute. Moreover, the training process is highly parallelizable as all optimal maps are independent and can thus be learned simultaneously. Finally, the approach is based solely on numerical linear algebra rather than minimizing a nonconvex objective function, allowing us to easily analyze and control the algorithm. We present several numerical experiments on both synthetic and real-world datasets to demonstrate the efficiency of our method. In particular, these experiments show that the proposed approach is highly competitive compared with state-of-the-art normalizing flows conditioned on time across a wide range of dimensionalities.
Authors: Seonghwan Kim, Jeheon Woo, Woo Youn Kim
The exploration of transition state (TS) geometries is crucial for elucidating chemical reaction mechanisms and modeling their kinetics. Recently, machine learning (ML) models have shown remarkable performance for prediction of TS geometries. However, they require 3D conformations of reactants and products often with their appropriate orientations as input, which demands substantial efforts and computational cost. Here, we propose a generative approach based on the stochastic diffusion method, namely TSDiff, for prediction of TS geometries just from 2D molecular graphs. TSDiff outperformed the existing ML models with 3D geometries in terms of both accuracy and efficiency. Moreover, it enables to sample various TS conformations, because it learned the distribution of TS geometries for diverse reactions in training. Thus, TSDiff was able to find more favorable reaction pathways with lower barrier heights than those in the reference database. These results demonstrate that TSDiff shows promising potential for an efficient and reliable TS exploration.
Authors: Steve Hanneke, Samory Kpotufe, Yasaman Mahdaviyeh
Theoretical studies on transfer learning or domain adaptation have so far focused on situations with a known hypothesis class or model; however in practice, some amount of model selection is usually involved, often appearing under the umbrella term of hyperparameter-tuning: for example, one may think of the problem of tuning for the right neural network architecture towards a target task, while leveraging data from a related source task.
Now, in addition to the usual tradeoffs on approximation vs estimation errors involved in model selection, this problem brings in a new complexity term, namely, the transfer distance between source and target distributions, which is known to vary with the choice of hypothesis class.
We present a first study of this problem, focusing on classification; in particular, the analysis reveals some remarkable phenomena: adaptive rates, i.e., those achievable with no distributional information, can be arbitrarily slower than oracle rates, i.e., when given knowledge on distances.
Authors: Xiaopeng Zhao, Zhenlin An, Qingrui Pan, Lei Yang
Although Maxwell discovered the physical laws of electromagnetic waves 160 years ago, how to precisely model the propagation of an RF signal in an electrically large and complex environment remains a long-standing problem. The difficulty is in the complex interactions between the RF signal and the obstacles (e.g., reflection, diffraction, etc.). Inspired by the great success of using a neural network to describe the optical field in computer vision, we propose a neural radio-frequency radiance field, NeRF$^\textbf{2}$, which represents a continuous volumetric scene function that makes sense of an RF signal's propagation. Particularly, after training with a few signal measurements, NeRF$^\textbf{2}$ can tell how/what signal is received at any position when it knows the position of a transmitter. As a physical-layer neural network, NeRF$^\textbf{2}$ can take advantage of the learned statistic model plus the physical model of ray tracing to generate a synthetic dataset that meets the training demands of application-layer artificial neural networks (ANNs). Thus, we can boost the performance of ANNs by the proposed turbo-learning, which mixes the true and synthetic datasets to intensify the training. Our experiment results show that turbo-learning can enhance performance with an approximate 50% increase. We also demonstrate the power of NeRF$^\textbf{2}$ in the field of indoor localization and 5G MIMO.
Authors: Erica Weng, Hana Hoshino, Deva Ramanan, Kris Kitani
Multi-modal trajectory forecasting methods commonly evaluate using single-agent metrics (marginal metrics), such as minimum Average Displacement Error (ADE) and Final Displacement Error (FDE), which fail to capture joint performance of multiple interacting agents. Only focusing on marginal metrics can lead to unnatural predictions, such as colliding trajectories or diverging trajectories for people who are clearly walking together as a group. Consequently, methods optimized for marginal metrics lead to overly-optimistic estimations of performance, which is detrimental to progress in trajectory forecasting research. In response to the limitations of marginal metrics, we present the first comprehensive evaluation of state-of-the-art (SOTA) trajectory forecasting methods with respect to multi-agent metrics (joint metrics): JADE, JFDE, and collision rate. We demonstrate the importance of joint metrics as opposed to marginal metrics with quantitative evidence and qualitative examples drawn from the ETH / UCY and Stanford Drone datasets. We introduce a new loss function incorporating joint metrics that, when applied to a SOTA trajectory forecasting method, achieves a 7\% improvement in JADE / JFDE on the ETH / UCY datasets with respect to the previous SOTA. Our results also indicate that optimizing for joint metrics naturally leads to an improvement in interaction modeling, as evidenced by a 16\% decrease in mean collision rate on the ETH / UCY datasets with respect to the previous SOTA. Code is available at \texttt{\hyperlink{https://github.com/ericaweng/joint-metrics-matter}{github.com/ericaweng/joint-metrics-matter}}.
Authors: Pierre Marion
Neural ordinary differential equations (neural ODEs) are a popular family of continuous-depth deep learning models. In this work, we consider a large family of parameterized ODEs with continuous-in-time parameters, which include time-dependent neural ODEs. We derive a generalization bound for this class by a Lipschitz-based argument. By leveraging the analogy between neural ODEs and deep residual networks, our approach yields in particular a generalization bound for a class of deep residual networks. The bound involves the magnitude of the difference between successive weight matrices. We illustrate numerically how this quantity affects the generalization capability of neural networks.
Authors: Mhairi Dunion, Trevor McInroe, Kevin Sebastian Luck, Josiah P. Hanna, Stefano V. Albrecht
Reinforcement Learning (RL) environments can produce training data with spurious correlations between features due to the amount of training data or its limited feature coverage. This can lead to RL agents encoding these misleading correlations in their latent representation, preventing the agent from generalising if the correlation changes within the environment or when deployed in the real world. Disentangled representations can improve robustness, but existing disentanglement techniques that minimise mutual information between features require independent features, thus they cannot disentangle correlated features. We propose an auxiliary task for RL algorithms that learns a disentangled representation of high-dimensional observations with correlated features by minimising the conditional mutual information between features in the representation. We demonstrate experimentally, using continuous control tasks, that our approach improves generalisation under correlation shifts, as well as improving the training performance of RL algorithms in the presence of correlated features.
Authors: Qingyun Wang, Doug Downey, Heng Ji, Tom Hope
Literature-Based Discovery (LBD) aims to discover new scientific knowledge by mining papers and generating hypotheses. Standard LBD is limited to predicting pairwise relations between discrete concepts (e.g., drug-disease links), and ignores critical contexts like experimental settings (e.g., a specific patient population where a drug is evaluated) and background motivations (e.g., to find drugs without specific side effects). We address these limitations with a novel formulation of contextualized-LBD (C-LBD): generating scientific hypotheses in natural language, while grounding them in a context that controls the hypothesis search space. We present a modeling framework using retrieval of ``inspirations'' from past scientific papers. Our evaluations reveal that GPT-4 tends to generate ideas with overall low technical depth and novelty, while our inspiration prompting approaches partially mitigate this issue. Our work represents a first step toward building language models that generate new ideas derived from scientific literature.
Authors: Daniël Vos, Jelle Vos, Tianyu Li, Zekeriya Erkin, Sicco Verwer
Decision trees are interpretable models that are well-suited to non-linear learning problems. Much work has been done on extending decision tree learning algorithms with differential privacy, a system that guarantees the privacy of samples within the training data. However, current state-of-the-art algorithms for this purpose sacrifice much utility for a small privacy benefit. These solutions create random decision nodes that reduce decision tree accuracy or spend an excessive share of the privacy budget on labeling leaves. Moreover, many works do not support continuous features or leak information about them. We propose a new method called PrivaTree based on private histograms that chooses good splits while consuming a small privacy budget. The resulting trees provide a significantly better privacy-utility trade-off and accept mixed numerical and categorical data without leaking information about numerical features. Finally, while it is notoriously hard to give robustness guarantees against data poisoning attacks, we demonstrate bounds for the expected accuracy and success rates of backdoor attacks against differentially-private learners. By leveraging the better privacy-utility trade-off of PrivaTree we are able to train decision trees with significantly better robustness against backdoor attacks compared to regular decision trees and with meaningful theoretical guarantees.
Authors: Sebastián Maldonado, Carla Vairetti, Katherine Jara, Miguel Carrasco, Julio López
In this paper, we propose a fuzzy adaptive loss function for enhancing deep learning performance in classification tasks. Specifically, we redefine the cross-entropy loss to effectively address class-level noise conditions, including the challenging problem of class imbalance. Our approach introduces aggregation operators, leveraging the power of fuzzy logic to improve classification accuracy. The rationale behind our proposed method lies in the iterative up-weighting of class-level components within the loss function, focusing on those with larger errors. To achieve this, we employ the ordered weighted average (OWA) operator and combine it with an adaptive scheme for gradient-based learning. Through extensive experimentation, our method outperforms other commonly used loss functions, such as the standard cross-entropy or focal loss, across various binary and multiclass classification tasks. Furthermore, we explore the influence of hyperparameters associated with the OWA operators and present a default configuration that performs well across different experimental settings.
Authors: Anthony Bardou, Patrick Thiran, Thomas Begin
Bayesian Optimization (BO) is typically used to optimize an unknown function $f$ that is noisy and costly to evaluate, by exploiting an acquisition function that must be maximized at each optimization step. Even if provably asymptotically optimal BO algorithms are efficient at optimizing low-dimensional functions, scaling them to high-dimensional spaces remains an open problem, often tackled by assuming an additive structure for $f$. By doing so, BO algorithms typically introduce additional restrictive assumptions on the additive structure that reduce their applicability domain. This paper contains two main contributions: (i) we relax the restrictive assumptions on the additive structure of $f$, at the expense of weakening the maximization guarantees of the acquisition function, and (ii) we address the over-exploration problem for decentralized BO algorithms. To these ends, we propose DumBO, an asymptotically optimal decentralized BO algorithm that achieves very competitive performance against state-of-the-art BO algorithms, especially when the additive structure of $f$ comprises high-dimensional factors.
Authors: Sadegh Mahdavi, Renjie Liao, Christos Thrampoulidis
Transformers have become the go-to architecture for language and vision tasks, yet their theoretical properties, especially memorization capacity, remain elusive. This paper investigates the memorization abilities of multi-head attention mechanisms, examining how many example sequences they can memorize, as a function of the number of heads and sequence length. Motivated by experimental findings on vision transformers, we introduce novel assumptions about the linear independence of input data, distinct from the commonly used general-position assumption. Under these assumptions, we demonstrate that an attention layer with $H$ heads, dimension $d$, and context size $n < d$, featuring $\Theta(Hd^2)$ parameters, can memorize $\Omega(Hn)$ examples. Our analysis sheds light on how different attention heads handle various example sequences, aided by the softmax operator's saturation property. We validate our findings through experiments on synthetic data.
Authors: Jayadeep Jacob, Tirthankar Bandyopadhyay, Jason Williams, Paulo Borges, Fabio Ramos
We propose to use a simulation driven inverse inference approach to model the dynamics of tree branches under manipulation. Learning branch dynamics and gaining the ability to manipulate deformable vegetation can help with occlusion-prone tasks, such as fruit picking in dense foliage, as well as moving overhanging vines and branches for navigation in dense vegetation. The underlying deformable tree geometry is encapsulated as coarse spring abstractions executed on parallel, non-differentiable simulators. The implicit statistical model defined by the simulator, reference trajectories obtained by actively probing the ground truth, and the Bayesian formalism, together guide the spring parameter posterior density estimation. Our non-parametric inference algorithm, based on Stein Variational Gradient Descent, incorporates biologically motivated assumptions into the inference process as neural network driven learnt joint priors; moreover, it leverages the finite difference scheme for gradient approximations. Real and simulated experiments confirm that our model can predict deformation trajectories, quantify the estimation uncertainty, and it can perform better when base-lined against other inference algorithms, particularly from the Monte Carlo family. The model displays strong robustness properties in the presence of heteroscedastic sensor noise; furthermore, it can generalise to unseen grasp locations.
Authors: Jiali Cui, Ying Nian Wu, Tian Han
This paper studies the fundamental problem of learning multi-layer generator models. The multi-layer generator model builds multiple layers of latent variables as a prior model on top of the generator, which benefits learning complex data distribution and hierarchical representations. However, such a prior model usually focuses on modeling inter-layer relations between latent variables by assuming non-informative (conditional) Gaussian distributions, which can be limited in model expressivity. To tackle this issue and learn more expressive prior models, we propose an energy-based model (EBM) on the joint latent space over all layers of latent variables with the multi-layer generator as its backbone. Such joint latent space EBM prior model captures the intra-layer contextual relations at each layer through layer-wise energy terms, and latent variables across different layers are jointly corrected. We develop a joint training scheme via maximum likelihood estimation (MLE), which involves Markov Chain Monte Carlo (MCMC) sampling for both prior and posterior distributions of the latent variables from different layers. To ensure efficient inference and learning, we further propose a variational training scheme where an inference model is used to amortize the costly posterior MCMC sampling. Our experiments demonstrate that the learned model can be expressive in generating high-quality images and capturing hierarchical features for better outlier detection.
Authors: Ziyan Wang, Hao Wang
Existing regression models tend to fall short in both accuracy and uncertainty estimation when the label distribution is imbalanced. In this paper, we propose a probabilistic deep learning model, dubbed variational imbalanced regression (VIR), which not only performs well in imbalanced regression but naturally produces reasonable uncertainty estimation as a byproduct. Different from typical variational autoencoders assuming I.I.D. representations (a data point's representation is not directly affected by other data points), our VIR borrows data with similar regression labels to compute the latent representation's variational distribution; furthermore, different from deterministic regression models producing point estimates, VIR predicts the entire normal-inverse-gamma distributions and modulates the associated conjugate distributions to impose probabilistic reweighting on the imbalanced data, thereby providing better uncertainty estimation. Experiments in several real-world datasets show that our VIR can outperform state-of-the-art imbalanced regression models in terms of both accuracy and uncertainty estimation. Code will soon be available at \url{https://github.com/Wang-ML-Lab/variational-imbalanced-regression}.
Authors: Ming Shi, Yingbin Liang, Ness Shroff
Partially observable Markov decision processes (POMDPs) have been widely applied to capture many real-world applications. However, existing theoretical results have shown that learning in general POMDPs could be intractable, where the main challenge lies in the lack of latent state information. A key fundamental question here is how much online state information (OSI) is sufficient to achieve tractability. In this paper, we establish a lower bound that reveals a surprising hardness result: unless we have full OSI, we need an exponentially scaling sample complexity to obtain an $\epsilon$-optimal policy solution for POMDPs. Nonetheless, inspired by the key insights in our lower bound design, we find that there exist important tractable classes of POMDPs even with only partial OSI. In particular, for two novel classes of POMDPs with partial OSI, we provide new algorithms that are proved to be near-optimal by establishing new regret upper and lower bounds.
Authors: Oliver Hoidn, Aashwin Ananda Mishra, Apurva Mehta
By circumventing the resolution limitations of optics, coherent diffractive imaging (CDI) and ptychography are making their way into scientific fields ranging from X-ray imaging to astronomy. Yet, the need for time consuming iterative phase recovery hampers real-time imaging. While supervised deep learning strategies have increased reconstruction speed, they sacrifice image quality. Furthermore, these methods' demand for extensive labeled training data is experimentally burdensome. Here, we propose an unsupervised physics-informed neural network reconstruction method, PtychoPINN, that retains the factor of 100-to-1000 speedup of deep learning-based reconstruction while improving reconstruction quality by combining the diffraction forward map with real-space constraints from overlapping measurements. In particular, PtychoPINN significantly advances generalizability, accuracy (with a typical 10 dB PSNR increase), and linear resolution (2- to 6-fold gain). This blend of performance and speed offers exciting prospects for high-resolution real-time imaging in high-throughput environments such as X-ray free electron lasers (XFELs) and diffraction-limited light sources.
Authors: Dante Everaert, Christopher Potts
It is often advantageous to train models on a subset of the available train examples, because the examples are of variable quality or because one would like to train with fewer examples, without sacrificing performance. We present Gradient Information Optimization (GIO), a scalable, task-agnostic approach to this data selection problem that requires only a small set of (unlabeled) examples representing a target distribution. GIO begins from a natural, information-theoretic objective that is intractable in practice. Our contribution is in showing that it can be made highly scalable through a simple relaxation of the objective and a highly efficient implementation. In experiments with machine translation, spelling correction, and image recognition, we show that GIO delivers outstanding results with very small train sets. These findings are robust to different representation models and hyperparameters for GIO itself. GIO is task- and domain-agnostic and can be applied out-of-the-box to new datasets and domains.
Authors: Zhenyuan Liu, Bart P. G. Van Parys, Henry Lam
In data-driven optimization, sample average approximation (SAA) is known to suffer from the so-called optimizer's curse that causes an over-optimistic evaluation of the solution performance. We argue that a special type of distributionallly robust optimization (DRO) formulation offers theoretical advantages in correcting for this optimizer's curse compared to simple ``margin'' adjustments to SAA and other DRO approaches: It attains a statistical bound on the out-of-sample performance, for a wide class of objective functions and distributions, that is nearly tightest in terms of exponential decay rate. This DRO uses an ambiguity set based on a Kullback Leibler (KL) divergence smoothed by the Wasserstein or L\'evy-Prokhorov (LP) distance via a suitable distance optimization. Computationally, we also show that such a DRO, and its generalized versions using smoothed $f$-divergence, are not harder than DRO problems based on $f$-divergence or Wasserstein distances, rendering our DRO formulations both statistically optimal and computationally viable.
Authors: Styfen Schär, Stefano Marelli, Bruno Sudret
We propose a novel surrogate modelling approach to efficiently and accurately approximate the response of complex dynamical systems driven by time-varying exogenous excitations over extended time periods. Our approach, namely manifold nonlinear autoregressive modelling with exogenous input (mNARX), involves constructing a problem-specific exogenous input manifold that is optimal for constructing autoregressive surrogates. The manifold, which forms the core of mNARX, is constructed incrementally by incorporating the physics of the system, as well as prior expert- and domain- knowledge. Because mNARX decomposes the full problem into a series of smaller sub-problems, each with a lower complexity than the original, it scales well with the complexity of the problem, both in terms of training and evaluation costs of the final surrogate. Furthermore, mNARX synergizes well with traditional dimensionality reduction techniques, making it highly suitable for modelling dynamical systems with high-dimensional exogenous inputs, a class of problems that is typically challenging to solve. Since domain knowledge is particularly abundant in physical systems, such as those found in civil and mechanical engineering, mNARX is well suited for these applications. We demonstrate that mNARX outperforms traditional autoregressive surrogates in predicting the response of a classical coupled spring-mass system excited by a one-dimensional random excitation. Additionally, we show that mNARX is well suited for emulating very high-dimensional time- and state-dependent systems, even when affected by active controllers, by surrogating the dynamics of a realistic aero-servo-elastic onshore wind turbine simulator. In general, our results demonstrate that mNARX offers promising prospects for modelling complex dynamical systems, in terms of accuracy and efficiency.
Authors: Yueh-Hua Wu, Xiaolong Wang, Masashi Hamaya
This paper introduces Elastic Decision Transformer (EDT), a significant advancement over the existing Decision Transformer (DT) and its variants. Although DT purports to generate an optimal trajectory, empirical evidence suggests it struggles with trajectory stitching, a process involving the generation of an optimal or near-optimal trajectory from the best parts of a set of sub-optimal trajectories. The proposed EDT differentiates itself by facilitating trajectory stitching during action inference at test time, achieved by adjusting the history length maintained in DT. Further, the EDT optimizes the trajectory by retaining a longer history when the previous trajectory is optimal and a shorter one when it is sub-optimal, enabling it to "stitch" with a more optimal trajectory. Extensive experimentation demonstrates EDT's ability to bridge the performance gap between DT-based and Q Learning-based approaches. In particular, the EDT outperforms Q Learning-based methods in a multi-task regime on the D4RL locomotion benchmark and Atari games. Videos are available at: https://kristery.github.io/edt/
Authors: Xuanlin Li, Yunhao Fang, Minghua Liu, Zhan Ling, Zhuowen Tu, Hao Su
Large vision-language models have achieved outstanding performance, but their size and computational requirements make their deployment on resource-constrained devices and time-sensitive tasks impractical. Model distillation, the process of creating smaller, faster models that maintain the performance of larger models, is a promising direction towards the solution. This paper investigates the distillation of visual representations in large teacher vision-language models into lightweight student models using a small- or mid-scale dataset. Notably, this study focuses on open-vocabulary out-of-distribution (OOD) generalization, a challenging problem that has been overlooked in previous model distillation literature. We propose two principles from vision and language modality perspectives to enhance student's OOD generalization: (1) by better imitating teacher's visual representation space, and carefully promoting better coherence in vision-language alignment with the teacher; (2) by enriching the teacher's language representations with informative and finegrained semantic attributes to effectively distinguish between different labels. We propose several metrics and conduct extensive experiments to investigate their techniques. The results demonstrate significant improvements in zero-shot and few-shot student performance on open-vocabulary out-of-distribution classification, highlighting the effectiveness of our proposed approaches. Poster: https://xuanlinli17.github.io/pdfs/iccv23_large_vlm_distillation_poster.pdf Code: https://github.com/xuanlinli17/large_vlm_distillation_ood
Authors: Seungyong Moon, Junyoung Yeom, Bumsoo Park, Hyun Oh Song
Discovering achievements with a hierarchical structure in procedurally generated environments presents a significant challenge. This requires an agent to possess a broad range of abilities, including generalization and long-term reasoning. Many prior methods have been built upon model-based or hierarchical approaches, with the belief that an explicit module for long-term planning would be advantageous for learning hierarchical dependencies. However, these methods demand an excessive number of environment interactions or large model sizes, limiting their practicality. In this work, we demonstrate that proximal policy optimization (PPO), a simple yet versatile model-free algorithm, outperforms previous methods when optimized with recent implementation practices. Moreover, we find that the PPO agent can predict the next achievement to be unlocked to some extent, albeit with limited confidence. Based on this observation, we introduce a novel contrastive learning method, called achievement distillation, which strengthens the agent's ability to predict the next achievement. Our method exhibits a strong capacity for discovering hierarchical achievements and shows state-of-the-art performance on the challenging Crafter environment in a sample-efficient manner while utilizing fewer model parameters.
Authors: Xiaohong Fan, Yin Yang, Ke Chen, Yujie Feng, Jianping Zhang
Proximal gradient-based optimization is one of the most common strategies to solve inverse problem of images, and it is easy to implement. However, these techniques often generate heavy artifacts in image reconstruction. One of the most popular refinement methods is to fine-tune the regularization parameter to alleviate such artifacts, but it may not always be sufficient or applicable due to increased computational costs. In this work, we propose a deep geometric incremental learning framework based on the second Nesterov proximal gradient optimization. The proposed end-to-end network not only has the powerful learning ability for high-/low-frequency image features, but also can theoretically guarantee that geometric texture details will be reconstructed from preliminary linear reconstruction. Furthermore, it can avoid the risk of intermediate reconstruction results falling outside the geometric decomposition domains and achieve fast convergence. Our reconstruction framework is decomposed into four modules including general linear reconstruction, cascade geometric incremental restoration, Nesterov acceleration, and post-processing. In the image restoration step, a cascade geometric incremental learning module is designed to compensate for missing texture information from different geometric spectral decomposition domains. Inspired by the overlap-tile strategy, we also develop a post-processing module to remove the block effect in patch-wise-based natural image reconstruction. All parameters in the proposed model are learnable, an adaptive initialization technique of physical parameters is also employed to make model flexibility and ensure converging smoothly. We compare the reconstruction performance of the proposed method with existing state-of-the-art methods to demonstrate its superiority. Our source codes are available at https://github.com/fanxiaohong/Nest-DGIL.
Authors: Jason Liang, Hormoz Shahrzad, Risto Miikkulainen
Many evolutionary algorithms (EAs) take advantage of parallel evaluation of candidates. However, if evaluation times vary significantly, many worker nodes (i.e.,\ compute clients) are idle much of the time, waiting for the next generation to be created. Evolutionary neural architecture search (ENAS), a class of EAs that optimizes the architecture and hyperparameters of deep neural networks, is particularly vulnerable to this issue. This paper proposes a generic asynchronous evaluation strategy (AES) that is then adapted to work with ENAS. AES increases throughput by maintaining a queue of up to $K$ individuals ready to be sent to the workers for evaluation and proceeding to the next generation as soon as $M<<K$ individuals have been evaluated. A suitable value for $M$ is determined experimentally, balancing diversity and efficiency. To showcase the generality and power of AES, it was first evaluated in eight-line sorting network design (a single-population optimization task with limited evaluation-time variability), achieving an over two-fold speedup. Next, it was evaluated in 11-bit multiplexer design (a single-population discovery task with extended variability), where a 14-fold speedup was observed. It was then scaled up to ENAS for image captioning (a multi-population open-ended-optimization task), resulting in an over two-fold speedup. In all problems, a multifold performance improvement was observed, suggesting that AES is a promising method for parallelizing the evolution of complex systems with long and variable evaluation times, such as those in ENAS.
Authors: Omer Veysel Cagatan, Baris Akgun
This paper introduces BarlowRL, a data-efficient reinforcement learning agent that combines the Barlow Twins self-supervised learning framework with DER (Data-Efficient Rainbow) algorithm. BarlowRL outperforms both DER and its contrastive counterpart CURL on the Atari 100k benchmark. BarlowRL avoids dimensional collapse by enforcing information spread to the whole space. This helps RL algorithms to utilize uniformly spread state representation that eventually results in a remarkable performance. The integration of Barlow Twins with DER enhances data efficiency and achieves superior performance in the RL tasks. BarlowRL demonstrates the potential of incorporating self-supervised learning techniques to improve RL algorithms.
Authors: Ching Chang, Wen-Chih Peng, Tien-Fu Chen
In this work, we leverage pre-trained Large Language Models (LLMs) to enhance time-series forecasting. Mirroring the growing interest in unifying models for Natural Language Processing and Computer Vision, we envision creating an analogous model for long-term time-series forecasting. Due to limited large-scale time-series data for building robust foundation models, our approach LLM4TS focuses on leveraging the strengths of pre-trained LLMs. By combining time-series patching with temporal encoding, we have enhanced the capability of LLMs to handle time-series data effectively. Inspired by the supervised fine-tuning in chatbot domains, we prioritize a two-stage fine-tuning process: first conducting supervised fine-tuning to orient the LLM towards time-series data, followed by task-specific downstream fine-tuning. Furthermore, to unlock the flexibility of pre-trained LLMs without extensive parameter adjustments, we adopt several Parameter-Efficient Fine-Tuning (PEFT) techniques. Drawing on these innovations, LLM4TS has yielded state-of-the-art results in long-term forecasting. Our model has also shown exceptional capabilities as both a robust representation learner and an effective few-shot learner, thanks to the knowledge transferred from the pre-trained LLM.
Authors: S. S. Hotegni, S. Peitz, M. Berkemeier
Different conflicting optimization criteria arise naturally in various Deep Learning scenarios. These can address different main tasks (i.e., in the setting of Multi-Task Learning), but also main and secondary tasks such as loss minimization versus sparsity. The usual approach is a simple weighting of the criteria, which formally only works in the convex setting. In this paper, we present a Multi-Objective Optimization algorithm using a modified Weighted Chebyshev scalarization for training Deep Neural Networks (DNNs) with respect to several tasks. By employing this scalarization technique, the algorithm can identify all optimal solutions of the original problem while reducing its complexity to a sequence of single-objective problems. The simplified problems are then solved using an Augmented Lagrangian method, enabling the use of popular optimization techniques such as Adam and Stochastic Gradient Descent, while efficaciously handling constraints. Our work aims to address the (economical and also ecological) sustainability issue of DNN models, with a particular focus on Deep Multi-Task models, which are typically designed with a very large number of weights to perform equally well on multiple tasks. Through experiments conducted on two Machine Learning datasets, we demonstrate the possibility of adaptively sparsifying the model during training without significantly impacting its performance, if we are willing to apply task-specific adaptations to the network weights. Code is available at https://github.com/salomonhotegni/MDMTN.
Authors: Raffaele Galliera, Kristen Brent Venable, Matteo Bassani, Niranjan Suri
In modern communication systems, efficient and reliable information dissemination is crucial for supporting critical operations across domains like disaster response, autonomous vehicles, and sensor networks. This paper introduces a Multi-Agent Reinforcement Learning (MARL) approach as a significant step forward in achieving more decentralized, efficient, and collaborative solutions. We propose a Partially Observable Stochastic Game (POSG) formulation for information dissemination empowering each agent to decide on message forwarding independently, based on their one-hop neighborhood. This constitutes a significant paradigm shift from traditional heuristics based on Multi-Point Relay (MPR) selection. Our approach harnesses Graph Convolutional Reinforcement Learning, employing Graph Attention Networks (GAT) with dynamic attention to capture essential network features. We propose two approaches, L-DGN and HL-DGN, which differ in the information that is exchanged among agents. We evaluate the performance of our decentralized approaches, by comparing them with a widely-used MPR heuristic, and we show that our trained policies are able to efficiently cover the network while bypassing the MPR set selection process. Our approach is a first step toward supporting the resilience of real-world broadcast communication infrastructures via learned, collaborative information dissemination.
Authors: Nazmus Sakib Ahmed, Saad Sakib Noor, Ashraful Islam Shanto Sikder, Abhijit Paul
This paper focuses on enhancing Bengali Document Layout Analysis (DLA) using the YOLOv8 model and innovative post-processing techniques. We tackle challenges unique to the complex Bengali script by employing data augmentation for model robustness. After meticulous validation set evaluation, we fine-tune our approach on the complete dataset, leading to a two-stage prediction strategy for accurate element segmentation. Our ensemble model, combined with post-processing, outperforms individual base architectures, addressing issues identified in the BaDLAD dataset. By leveraging this approach, we aim to advance Bengali document analysis, contributing to improved OCR and document comprehension and BaDLAD serves as a foundational resource for this endeavor, aiding future research in the field. Furthermore, our experiments provided key insights to incorporate new strategies into the established solution.
Authors: Yichong Leng, Zhifang Guo, Kai Shen, Xu Tan, Zeqian Ju, Yanqing Liu, Yufei Liu, Dongchao Yang, Leying Zhang, Kaitao Song, Lei He, Xiang-Yang Li, Sheng Zhao, Tao Qin, Jiang Bian
Speech conveys more information than text, as the same word can be uttered in various voices to convey diverse information. Compared to traditional text-to-speech (TTS) methods relying on speech prompts (reference speech) for voice variability, using text prompts (descriptions) is more user-friendly since speech prompts can be hard to find or may not exist at all. TTS approaches based on the text prompt face two main challenges: 1) the one-to-many problem, where not all details about voice variability can be described in the text prompt, and 2) the limited availability of text prompt datasets, where vendors and large cost of data labeling are required to write text prompts for speech. In this work, we introduce PromptTTS 2 to address these challenges with a variation network to provide variability information of voice not captured by text prompts, and a prompt generation pipeline to utilize the large language models (LLM) to compose high quality text prompts. Specifically, the variation network predicts the representation extracted from the reference speech (which contains full information about voice variability) based on the text prompt representation. For the prompt generation pipeline, it generates text prompts for speech with a speech language understanding model to recognize voice attributes (e.g., gender, speed) from speech and a large language model to formulate text prompts based on the recognition results. Experiments on a large-scale (44K hours) speech dataset demonstrate that compared to the previous works, PromptTTS 2 generates voices more consistent with text prompts and supports the sampling of diverse voice variability, thereby offering users more choices on voice generation. Additionally, the prompt generation pipeline produces high-quality text prompts, eliminating the large labeling cost. The demo page of PromptTTS 2 is available online.
Authors: Ze Peng, Lei Qi, Yinghuan Shi, Yang Gao
A recent empirical observation (Li et al., 2022b) of activation sparsity in MLP blocks offers an opportunity to drastically reduce computation costs for free. Although having attributed it to training dynamics, existing theoretical explanations of activation sparsity are restricted to shallow networks, small training steps and special training, despite its emergence in deep models standardly trained for a large number of steps. To fill these gaps, we propose the notion of gradient sparsity as one source of activation sparsity and a theoretical explanation based on it that sees sparsity a necessary step to adversarial robustness w.r.t. hidden features and parameters, which is approximately the flatness of minima for well-learned models. The theory applies to standardly trained LayerNorm-ed MLPs, and further to Transformers or other architectures trained with weight noises. Eliminating other sources of flatness except for sparsity, we discover the phenomenon that the ratio between the largest and smallest non-zero singular values of weight matrices is small. When discussing the emergence of this spectral concentration, we use random matrix theory (RMT) as a powerful tool to analyze stochastic gradient noises. Validational experiments are conducted to verify our gradient-sparsity-based explanation. We propose two plug-and-play modules for both training and finetuning for sparsity. Experiments on ImageNet-1k and C4 demonstrate their 50% sparsity improvements, indicating further potential cost reduction in both training and inference.
Authors: Ju Qi, Ting Feng, Falun Hei, Zhemei Fang, Yunfeng Luo
Counterfactual Regret Minimization (CFR) and its variants are the best algorithms so far for solving large-scale incomplete information games. However, we believe that there are two problems with CFR: First, matrix multiplication is required in CFR iteration, and the time complexity of one iteration is too high; Secondly, the game characteristics in the real world are different. Just using one CFR algorithm will not be perfectly suitable for all game problems.
For these two problems, this paper proposes a new algorithm called Pure CFR (PCFR) based on CFR. PCFR can be seen as a combination of CFR and Fictitious Play (FP), inheriting the concept of counterfactual regret (value) from CFR, and using the best response strategy instead of the regret matching strategy for the next iteration. This algorithm has three advantages. First, PCFR can be combined with any CFR variant. The resulting Pure MCCFR (PMCCFR) can significantly reduce the time and space complexity of one iteration. Secondly, our experiments show that the convergence speed of the PMCCFR is 2$\sim$3 times that of the MCCFR. Finally, there is a type of game that is very suitable for PCFR, we call this type of game clear-game, which is characterized by a high proportion of dominated strategies. Experiments show that in clear-game, the convergence rate of PMCCFR is two orders of magnitude higher than that of MCCFR.
Authors: David DeFazio, Eisuke Hirota, Shiqi Zhang
Seeing-eye robots are very useful tools for guiding visually impaired people, potentially producing a huge societal impact given the low availability and high cost of real guide dogs. Although a few seeing-eye robot systems have already been demonstrated, none considered external tugs from humans, which frequently occur in a real guide dog setting. In this paper, we simultaneously train a locomotion controller that is robust to external tugging forces via Reinforcement Learning (RL), and an external force estimator via supervised learning. The controller ensures stable walking, and the force estimator enables the robot to respond to the external forces from the human. These forces are used to guide the robot to the global goal, which is unknown to the robot, while the robot guides the human around nearby obstacles via a local planner. Experimental results in simulation and on hardware show that our controller is robust to external forces, and our seeing-eye system can accurately detect force direction. We demonstrate our full seeing-eye robot system on a real quadruped robot with a blindfolded human. The video can be seen at our project page: https://bu-air-lab.github.io/guide_dog/
Authors: Zhengxiang Shi, Aldo Lipani
Prompt tuning (PT), where a small amount of trainable soft (continuous) prompt vectors is affixed to the input of language models (LM), has shown promising results across various tasks and models for parameter-efficient fine-tuning (PEFT). PT stands out from other PEFT approaches because it maintains competitive performance with fewer trainable parameters and does not drastically scale up its parameters as the model size expands. However, PT introduces additional soft prompt tokens, leading to longer input sequences, which significantly impacts training and inference time and memory usage due to the Transformer's quadratic complexity. Particularly concerning for Large Language Models (LLMs) that face heavy daily querying. To address this issue, we propose Decomposed Prompt Tuning (DePT), which decomposes the soft prompt into a shorter soft prompt and a pair of low-rank matrices that are then optimised with two different learning rates. This allows DePT to achieve better performance while saving over 20% memory and time costs compared to vanilla PT and its variants, without changing trainable parameter sizes. Through extensive experiments on 23 natural language processing (NLP) and vision-language (VL) tasks, we demonstrate that DePT outperforms state-of-the-art PEFT approaches, including the full fine-tuning baseline in some scenarios. Additionally, we empirically show that DEPT grows more efficient as the model size increases. Our further study reveals that DePT integrates seamlessly with parameter-efficient transfer learning in the few-shot learning setting and highlights its adaptability to various model architectures and sizes.
Authors: Mengyuan Zhang, Kai Liu
Sparse logistic regression is for classification and feature selection simultaneously. Although many studies have been done to solve $\ell_1$-regularized logistic regression, there is no equivalently abundant work on solving sparse logistic regression with nonconvex regularization term. In this paper, we propose a unified framework to solve $\ell_1$-regularized logistic regression, which can be naturally extended to nonconvex regularization term, as long as certain requirement is satisfied. In addition, we also utilize a different line search criteria to guarantee monotone convergence for various regularization terms. Empirical experiments on binary classification tasks with real-world datasets demonstrate our proposed algorithms are capable of performing classification and feature selection effectively at a lower computational cost.
Authors: Xingyao Wang, Zihan Wang, Jiateng Liu, Yangyi Chen, Lifan Yuan, Hao Peng, Heng Ji
To solve complex tasks, large language models (LLMs) often require multiple rounds of interactions with the user, sometimes assisted by external tools. However, current evaluation protocols often emphasize benchmark performance with single-turn exchanges, neglecting the nuanced interactions among the user, LLMs, and external tools, while also underestimating the importance of natural language feedback from users. These oversights contribute to discrepancies between research benchmark evaluations and real-world use cases. We introduce MINT, a benchmark that evaluates LLMs' ability to solve tasks with multi-turn interactions by (1) using tools and (2) leveraging natural language feedback. To ensure reproducibility, we provide an evaluation framework where LLMs can access tools by executing Python code and receive users' natural language feedback simulated by GPT-4. We repurpose a diverse set of established evaluation datasets focusing on reasoning, coding, and decision-making and carefully curate them into a compact subset for efficient evaluation. Our analysis of 20 open- and closed-source LLMs offers intriguing findings. (a) LLMs generally benefit from tools and language feedback, with performance gains (absolute, same below) of 1-8% for each turn of tool use and 2-17% with natural language feedback. (b) Better single-turn performance does not guarantee better multi-turn performance. (c) Surprisingly, on the LLMs evaluated, supervised instruction-finetuning (SIFT) and reinforcement learning from human feedback (RLHF) generally hurt multi-turn capabilities. We expect MINT can help measure progress and incentivize research in improving LLMs' capabilities in multi-turn interactions, especially for open-source communities where multi-turn human evaluation can be less accessible compared to commercial LLMs with a larger user base.
Authors: Zihan Zhou, Honghao Wei, Lei Ying
This paper considers the best policy identification (BPI) problem in online Constrained Markov Decision Processes (CMDPs). We are interested in algorithms that are model-free, have low regret, and identify an optimal policy with a high probability. Existing model-free algorithms for online CMDPs with sublinear regret and constraint violation do not provide any convergence guarantee to an optimal policy and provide only average performance guarantees when a policy is uniformly sampled at random from all previously used policies. In this paper, we develop a new algorithm, named Pruning-Refinement-Identification (PRI), based on a fundamental structural property of CMDPs we discover, called limited stochasticity. The property says for a CMDP with $N$ constraints, there exists an optimal policy with at most $N$ stochastic decisions.
The proposed algorithm first identifies at which step and in which state a stochastic decision has to be taken and then fine-tunes the distributions of these stochastic decisions. PRI achieves trio objectives: (i) PRI is a model-free algorithm; and (ii) it outputs a near-optimal policy with a high probability at the end of learning; and (iii) in the tabular setting, PRI guarantees $\tilde{\mathcal{O}}(\sqrt{K})$ regret and constraint violation, which significantly improves the best existing regret bound $\tilde{\mathcal{O}}(K^{\frac{4}{5}})$ under a model-free algorithm, where $K$ is the total number of episodes.
Authors: Fabian Mentzer, David Minnen, Eirikur Agustsson, Michael Tschannen
We propose to replace vector quantization (VQ) in the latent representation of VQ-VAEs with a simple scheme termed finite scalar quantization (FSQ), where we project the VAE representation down to a few dimensions (typically less than 10). Each dimension is quantized to a small set of fixed values, leading to an (implicit) codebook given by the product of these sets. By appropriately choosing the number of dimensions and values each dimension can take, we obtain the same codebook size as in VQ. On top of such discrete representations, we can train the same models that have been trained on VQ-VAE representations. For example, autoregressive and masked transformer models for image generation, multimodal generation, and dense prediction computer vision tasks. Concretely, we employ FSQ with MaskGIT for image generation, and with UViM for depth estimation, colorization, and panoptic segmentation. Despite the much simpler design of FSQ, we obtain competitive performance in all these tasks. We emphasize that FSQ does not suffer from codebook collapse and does not need the complex machinery employed in VQ (commitment losses, codebook reseeding, code splitting, entropy penalties, etc.) to learn expressive discrete representations.
Authors: David Jin, Niclas Kannengießer, Sascha Rank, Ali Sunyaev
To leverage data for the sufficient training of machine learning (ML) models from multiple parties in a confidentiality-preserving way, various collaborative distributed ML (CDML) system designs have been developed, for example, to perform assisted learning, federated learning, and split learning. CDML system designs show different traits, including high agent autonomy, ML model confidentiality, and fault tolerance. Facing a wide variety of CDML system designs with different traits, it is difficult for developers to design CDML systems with traits that match use case requirements in a targeted way. However, inappropriate CDML system designs may result in CDML systems failing their envisioned purposes. We developed a CDML design toolbox that can guide the development of CDML systems. Based on the CDML design toolbox, we present CDML system archetypes with distinct key traits that can support the design of CDML systems to meet use case requirements.
Authors: Kai Weixian Lan, Elias Gueidon, Ayano Kaneda, Julian Panetta, Joseph Teran
We introduce a neural-preconditioned iterative solver for Poisson equations with mixed boundary conditions. The Poisson equation is ubiquitous in scientific computing: it governs a wide array of physical phenomena, arises as a subproblem in many numerical algorithms, and serves as a model problem for the broader class of elliptic PDEs. The most popular Poisson discretizations yield large sparse linear systems. At high resolution, and for performance-critical applications, iterative solvers can be advantageous for these -- but only when paired with powerful preconditioners. The core of our solver is a neural network trained to approximate the inverse of a discrete structured-grid Laplace operator for a domain of arbitrary shape and with mixed boundary conditions. The structure of this problem motivates a novel network architecture that we demonstrate is highly effective as a preconditioner even for boundary conditions outside the training set. We show that on challenging test cases arising from an incompressible fluid simulation, our method outperforms state-of-the-art solvers like algebraic multigrid as well as some recent neural preconditioners.
Authors: Sjoerd Dirksen, Patrick Finke, Martin Genzel
In practice, deep neural networks are often able to easily interpolate their training data. To understand this phenomenon, many works have aimed to quantify the memorization capacity of a neural network architecture: the largest number of points such that the architecture can interpolate any placement of these points with any assignment of labels. For real-world data, however, one intuitively expects the presence of a benign structure so that interpolation already occurs at a smaller network size than suggested by memorization capacity. In this paper, we investigate interpolation by adopting an instance-specific viewpoint. We introduce a simple randomized algorithm that, given a fixed finite dataset with two classes, with high probability constructs an interpolating three-layer neural network in polynomial time. The required number of parameters is linked to geometric properties of the two classes and their mutual arrangement. As a result, we obtain guarantees that are independent of the number of samples and hence move beyond worst-case memorization capacity bounds. We illustrate the effectiveness of the algorithm in non-pathological situations with extensive numerical experiments and link the insights back to the theoretical results.
Authors: KaiChieh Lo, Daniel Huang
We refer to the setting where the (partial) derivatives of a neural network's (NN's) predictions with respect to its inputs are used as additional training signal as a derivative-constrained (DC) NN. This situation is common in physics-informed settings in the natural sciences. We propose an integrated RELU (IReLU) activation function to improve training of DC NNs. We also investigate denormalization and label rescaling to help stabilize DC training. We evaluate our methods on physics-informed settings including quantum chemistry and Scientific Machine Learning (SciML) tasks. We demonstrate that existing architectures with IReLU activations combined with denormalization and label rescaling better incorporate training signal provided by derivative constraints.
Authors: Zhang-Wei Hong, Aviral Kumar, Sathwik Karnik, Abhishek Bhandwaldar, Akash Srivastava, Joni Pajarinen, Romain Laroche, Abhishek Gupta, Pulkit Agrawal
Offline policy learning is aimed at learning decision-making policies using existing datasets of trajectories without collecting additional data. The primary motivation for using reinforcement learning (RL) instead of supervised learning techniques such as behavior cloning is to find a policy that achieves a higher average return than the trajectories constituting the dataset. However, we empirically find that when a dataset is dominated by suboptimal trajectories, state-of-the-art offline RL algorithms do not substantially improve over the average return of trajectories in the dataset. We argue this is due to an assumption made by current offline RL algorithms of staying close to the trajectories in the dataset. If the dataset primarily consists of sub-optimal trajectories, this assumption forces the policy to mimic the suboptimal actions. We overcome this issue by proposing a sampling strategy that enables the policy to only be constrained to ``good data" rather than all actions in the dataset (i.e., uniform sampling). We present a realization of the sampling strategy and an algorithm that can be used as a plug-and-play module in standard offline RL algorithms. Our evaluation demonstrates significant performance gains in 72 imbalanced datasets, D4RL dataset, and across three different offline RL algorithms. Code is available at https://github.com/Improbable-AI/dw-offline-rl.
Authors: Shuaiwen Leon Song, Bonnie Kruft, Minjia Zhang, Conglong Li, Shiyang Chen, Chengming Zhang, Masahiro Tanaka, Xiaoxia Wu, Jeff Rasley, Ammar Ahmad Awan, Connor Holmes, Martin Cai, Adam Ghanem, Zhongzhu Zhou, Yuxiong He, Pete Luferenko, Divya Kumar, Jonathan Weyn, Ruixiong Zhang, Sylwester Klocek, Volodymyr Vragov, Mohammed AlQuraishi, Gustaf Ahdritz, Christina Floristean, Cristina Negri, Rao Kotamarthi, Venkatram Vishwanath, Arvind Ramanathan, Sam Foreman, Kyle Hippe, Troy Arcomano, Romit Maulik, Maxim Zvyagin, Alexander Brace, Bin Zhang, Cindy Orozco Bohorquez, Austin Clyde, Bharat Kale, Danilo Perez-Rivera, Heng Ma, Carla M. Mann, Michael Irvin, J. Gregory Pauloski, Logan Ward, Valerie Hayot, Murali Emani, Zhen Xie, Diangen Lin, Maulik Shukla, Ian Foster, James J. Davis, Michael E. Papka, et al. (40 additional authors not shown)
In the upcoming decade, deep learning may revolutionize the natural sciences, enhancing our capacity to model and predict natural occurrences. This could herald a new era of scientific exploration, bringing significant advancements across sectors from drug development to renewable energy. To answer this call, we present DeepSpeed4Science initiative (deepspeed4science.ai) which aims to build unique capabilities through AI system technology innovations to help domain experts to unlock today's biggest science mysteries. By leveraging DeepSpeed's current technology pillars (training, inference and compression) as base technology enablers, DeepSpeed4Science will create a new set of AI system technologies tailored for accelerating scientific discoveries by addressing their unique complexity beyond the common technical approaches used for accelerating generic large language models (LLMs). In this paper, we showcase the early progress we made with DeepSpeed4Science in addressing two of the critical system challenges in structural biology research.
Authors: Defu Cao, Furong Jia, Sercan O Arik, Tomas Pfister, Yixiang Zheng, Wen Ye, Yan Liu
The past decade has witnessed significant advances in time series modeling with deep learning. While achieving state-of-the-art results, the best-performing architectures vary highly across applications and domains. Meanwhile, for natural language processing, the Generative Pre-trained Transformer (GPT) has demonstrated impressive performance via training one general-purpose model across various textual datasets. It is intriguing to explore whether GPT-type architectures can be effective for time series, capturing the intrinsic dynamic attributes and leading to significant accuracy improvements. In this paper, we propose a novel framework, TEMPO, that can effectively learn time series representations. We focus on utilizing two essential inductive biases of the time series task for pre-trained models: (i) decomposition of the complex interaction between trend, seasonal and residual components; and (ii) introducing the selection-based prompts to facilitate distribution adaptation in non-stationary time series. TEMPO expands the capability for dynamically modeling real-world temporal phenomena from data within diverse domains. Our experiments demonstrate the superior performance of TEMPO over state-of-the-art methods on a number of time series benchmark datasets. This performance gain is observed not only in standard supervised learning settings but also in scenarios involving previously unseen datasets as well as in scenarios with multi-modal inputs. This compelling finding highlights TEMPO's potential to constitute a foundational model-building framework.
Authors: Katharine M. Clark, Paul D. McNicholas
Matrix-variate distributions are a recent addition to the model-based clustering field, thereby making it possible to analyze data in matrix form with complex structure such as images and time series. Due to its recent appearance, there is limited literature on matrix-variate data, with even less on dealing with outliers in these models. An approach for clustering matrix-variate normal data with outliers is discussed. The approach, which uses the distribution of subset log-likelihoods, extends the OCLUST algorithm to matrix-variate normal data and uses an iterative approach to detect and trim outliers.
Authors: Doyup Lee, Chiheon Kim, Minsu Cho, Wook-Shin Han
Generalizable implicit neural representation (INR) enables a single continuous function, i.e., a coordinate-based neural network, to represent multiple data instances by modulating its weights or intermediate features using latent codes. However, the expressive power of the state-of-the-art modulation is limited due to its inability to localize and capture fine-grained details of data entities such as specific pixels and rays. To address this issue, we propose a novel framework for generalizable INR that combines a transformer encoder with a locality-aware INR decoder. The transformer encoder predicts a set of latent tokens from a data instance to encode local information into each latent token. The locality-aware INR decoder extracts a modulation vector by selectively aggregating the latent tokens via cross-attention for a coordinate input and then predicts the output by progressively decoding with coarse-to-fine modulation through multiple frequency bandwidths. The selective token aggregation and the multi-band feature modulation enable us to learn locality-aware representation in spatial and spectral aspects, respectively. Our framework significantly outperforms previous generalizable INRs and validates the usefulness of the locality-aware latents for downstream tasks such as image generation.
Authors: Lizhang Chen, Bo Liu, Kaizhao Liang, Qiang Liu
Lion (Evolved Sign Momentum), a new optimizer discovered through program search, has shown promising results in training large AI models. It performs comparably or favorably to AdamW but with greater memory efficiency. As we can expect from the results of a random search program, Lion incorporates elements from several existing algorithms, including signed momentum, decoupled weight decay, Polak, and Nesterov momentum, but does not fit into any existing category of theoretically grounded optimizers. Thus, even though Lion appears to perform well as a general-purpose optimizer for a wide range of tasks, its theoretical basis remains uncertain. This lack of theoretical clarity limits opportunities to further enhance and expand Lion's efficacy.
This work aims to demystify Lion. Based on both continuous-time and discrete-time analysis, we demonstrate that Lion is a theoretically novel and principled approach for minimizing a general loss function $f(x)$ while enforcing a bound constraint $\|x\|_\infty \leq 1/\lambda$. Lion achieves this through the incorporation of decoupled weight decay, where $\lambda$ represents the weight decay coefficient. Our analysis is made possible by the development of a new Lyapunov function for the Lion updates. It applies to a broader family of Lion-$\kappa$ algorithms, where the $\text{sign}(\cdot)$ operator in Lion is replaced by the subgradient of a convex function $\kappa$, leading to the solution of a general composite optimization problem of $\min_x f(x) + \kappa^*(x)$. Our findings provide valuable insights into the dynamics of Lion and pave the way for further improvements and extensions of Lion-related algorithms.
Authors: Bruno Silva, Leonardo Nunes, Roberto Estevão, Vijay Aski, Ranveer Chandra
Large language models (LLMs) have demonstrated remarkable capabilities in natural language understanding across various domains, including healthcare and finance. For some tasks, LLMs achieve similar or better performance than trained human beings, therefore it is reasonable to employ human exams (e.g., certification tests) to assess the performance of LLMs. We present a comprehensive evaluation of popular LLMs, such as Llama 2 and GPT, on their ability to answer agriculture-related questions. In our evaluation, we also employ RAG (Retrieval-Augmented Generation) and ER (Ensemble Refinement) techniques, which combine information retrieval, generation capabilities, and prompting strategies to improve the LLMs' performance. To demonstrate the capabilities of LLMs, we selected agriculture exams and benchmark datasets from three of the largest agriculture producer countries: Brazil, India, and the USA. Our analysis highlights GPT-4's ability to achieve a passing score on exams to earn credits for renewing agronomist certifications, answering 93% of the questions correctly and outperforming earlier general-purpose models, which achieved 88% accuracy. On one of our experiments, GPT-4 obtained the highest performance when compared to human subjects. This performance suggests that GPT-4 could potentially pass on major graduate education admission tests or even earn credits for renewing agronomy certificates. We also explore the models' capacity to address general agriculture-related questions and generate crop management guidelines for Brazilian and Indian farmers, utilizing robust datasets from the Brazilian Agency of Agriculture (Embrapa) and graduate program exams from India. The results suggest that GPT-4, ER, and RAG can contribute meaningfully to agricultural education, assessment, and crop management practice, offering valuable insights to farmers and agricultural professionals.
Authors: Tianlong Li, Wenhao Liu, Changze Lv, Jianhan Xu, Cenyuan Zhang, Muling Wu, Xiaoqing Zheng, Xuanjing Huang
Spiking neural networks (SNNs) have demonstrated the capability to achieve comparable performance to deep neural networks (DNNs) in both visual and linguistic domains while offering the advantages of improved energy efficiency and adherence to biological plausibility. However, the extension of such single-modality SNNs into the realm of multimodal scenarios remains an unexplored territory. Drawing inspiration from the concept of contrastive language-image pre-training (CLIP), we introduce a novel framework, named SpikeCLIP, to address the gap between two modalities within the context of spike-based computing through a two-step recipe involving ``Alignment Pre-training + Dual-Loss Fine-tuning". Extensive experiments demonstrate that SNNs achieve comparable results to their DNN counterparts while significantly reducing energy consumption across a variety of datasets commonly used for multimodal model evaluation. Furthermore, SpikeCLIP maintains robust performance in image classification tasks that involve class labels not predefined within specific categories.
Authors: Qizhi Pei, Kaiyuan Gao, Lijun Wu, Jinhua Zhu, Yingce Xia, Shufang Xie, Tao Qin, Kun He, Tie-Yan Liu, Rui Yan
Modeling the interaction between proteins and ligands and accurately predicting their binding structures is a critical yet challenging task in drug discovery. Recent advancements in deep learning have shown promise in addressing this challenge, with sampling-based and regression-based methods emerging as two prominent approaches. However, these methods have notable limitations. Sampling-based methods often suffer from low efficiency due to the need for generating multiple candidate structures for selection. On the other hand, regression-based methods offer fast predictions but may experience decreased accuracy. Additionally, the variation in protein sizes often requires external modules for selecting suitable binding pockets, further impacting efficiency. In this work, we propose $\mathbf{FABind}$, an end-to-end model that combines pocket prediction and docking to achieve accurate and fast protein-ligand binding. $\mathbf{FABind}$ incorporates a unique ligand-informed pocket prediction module, which is also leveraged for docking pose estimation. The model further enhances the docking process by incrementally integrating the predicted pocket to optimize protein-ligand binding, reducing discrepancies between training and inference. Through extensive experiments on benchmark datasets, our proposed $\mathbf{FABind}$ demonstrates strong advantages in terms of effectiveness and efficiency compared to existing methods. Our code is available at $\href{https://github.com/QizhiPei/FABind}{Github}$.
Authors: Mouïn Ben Ammar, Nacim Belkhir, Sebastian Popescu, Antoine Manzanera, Gianni Franchi
Detecting out-of-distribution (OOD) data is a critical challenge in machine learning due to model overconfidence, often without awareness of their epistemological limits. We hypothesize that ``neural collapse'', a phenomenon affecting in-distribution data for models trained beyond loss convergence, also influences OOD data. To benefit from this interplay, we introduce NECO, a novel post-hoc method for OOD detection, which leverages the geometric properties of ``neural collapse'' and of principal component spaces to identify OOD data. Our extensive experiments demonstrate that NECO achieves state-of-the-art results on both small and large-scale OOD detection tasks while exhibiting strong generalization capabilities across different network architectures. Furthermore, we provide a theoretical explanation for the effectiveness of our method in OOD detection. We plan to release the code after the anonymity period.
Authors: Joël Mathys, Florian Grötschla, Kalyan Varma Nadimpalli, Roger Wattenhofer
Graph Neural Networks are a natural fit for learning algorithms. They can directly represent tasks through an abstract but versatile graph structure and handle inputs of different sizes. This opens up the possibility for scaling and extrapolation to larger graphs, one of the most important advantages of an algorithm. However, this raises two core questions i) How can we enable nodes to gather the required information in a given graph ($\textit{information exchange}$), even if is far away and ii) How can we design an execution framework which enables this information exchange for extrapolation to larger graph sizes ($\textit{algorithmic alignment for extrapolation}$). We propose a new execution framework that is inspired by the design principles of distributed algorithms: Flood and Echo Net. It propagates messages through the entire graph in a wave like activation pattern, which naturally generalizes to larger instances. Through its sparse but parallel activations it is provably more efficient in terms of message complexity. We study the proposed model and provide both empirical evidence and theoretical insights in terms of its expressiveness, efficiency, information exchange and ability to extrapolate.
Authors: Zheshun Wu, Zenglin Xu, Dun Zeng, Qifan Wang
Federated Learning (FL) has surged in prominence due to its capability of collaborative model training without direct data sharing. However, the vast disparity in local data distributions among clients, often termed the non-Independent Identically Distributed (non-IID) challenge, poses a significant hurdle to FL's generalization efficacy. The scenario becomes even more complex when not all clients participate in the training process, a common occurrence due to unstable network connections or limited computational capacities. This can greatly complicate the assessment of the trained models' generalization abilities. While a plethora of recent studies has centered on the generalization gap pertaining to unseen data from participating clients with diverse distributions, the divergence between the training distributions of participating clients and the testing distributions of non-participating ones has been largely overlooked. In response, our paper unveils an information-theoretic generalization framework for FL. Specifically, it quantifies generalization errors by evaluating the information entropy of local distributions and discerning discrepancies across these distributions. Inspired by our deduced generalization bounds, we introduce a weighted aggregation approach and a duo of client selection strategies. These innovations aim to bolster FL's generalization prowess by encompassing a more varied set of client data distributions. Our extensive empirical evaluations reaffirm the potency of our proposed methods, aligning seamlessly with our theoretical construct.
Authors: Huayu Chen, Cheng Lu, Zhengyi Wang, Hang Su, Jun Zhu
Recent developments in offline reinforcement learning have uncovered the immense potential of diffusion modeling, which excels at representing heterogeneous behavior policies. However, sampling from diffusion policies is considerably slow because it necessitates tens to hundreds of iterative inference steps for one action. To address this issue, we propose to extract an efficient deterministic inference policy from critic models and pretrained diffusion behavior models, leveraging the latter to directly regularize the policy gradient with the behavior distribution's score function during optimization. Our method enjoys powerful generative capabilities of diffusion modeling while completely circumventing the computationally intensive and time-consuming diffusion sampling scheme, both during training and evaluation. Extensive results on D4RL tasks show that our method boosts action sampling speed by more than 25 times compared with various leading diffusion-based methods in locomotion tasks, while still maintaining state-of-the-art performance.
Authors: Mehdi Letafati, Samad Ali, Matti Latva-aho
Innovative foundation models, such as GPT-3 and stable diffusion models, have made a paradigm shift in the realm of artificial intelligence (AI) towards generative AI-based systems. In unison, from data communication and networking perspective, AI and machine learning (AI/ML) algorithms are envisioned to be pervasively incorporated into the future generations of wireless communications systems, highlighting the need for novel AI-native solutions for the emergent communication scenarios. In this article, we outline the applications of generative AI in wireless communication systems to lay the foundations for research in this field. Diffusion-based generative models, as the new state-of-the-art paradigm of generative models, are introduced, and their applications in wireless communication systems are discussed. Two case studies are also presented to showcase how diffusion models can be exploited for the development of resilient AI-native communication systems. Specifically, we propose denoising diffusion probabilistic models (DDPM) for a wireless communication scheme with non-ideal transceivers, where 30% improvement is achieved in terms of bit error rate. As the second application, DDPMs are employed at the transmitter to shape the constellation symbols, highlighting a robust out-of-distribution performance. Finally, future directions and open issues for the development of generative AI-based wireless systems are discussed to promote future research endeavors towards wireless generative AI (WiGenAI).
Authors: Yun Zhu, Yaoke Wang, Haizhou Shi, Zhenshuo Zhang, Siliang Tang
Graph-structured data is ubiquitous in the world which models complex relationships between objects, enabling various Web applications. Daily influxes of unlabeled graph data on the Web offer immense potential for these applications. Graph self-supervised algorithms have achieved significant success in acquiring generic knowledge from abundant unlabeled graph data. These pre-trained models can be applied to various downstream Web applications, saving training time and improving downstream (target) performance. However, different graphs, even across seemingly similar domains, can differ significantly in terms of attribute semantics, posing difficulties, if not infeasibility, for transferring the pre-trained models to downstream tasks. Concretely speaking, for example, the additional task-specific node information in downstream tasks (specificity) is usually deliberately omitted so that the pre-trained representation (transferability) can be leveraged. The trade-off as such is termed as "transferability-specificity dilemma" in this work. To address this challenge, we introduce an innovative deployment module coined as GraphControl, motivated by ControlNet, to realize better graph domain transfer learning. Specifically, by leveraging universal structural pre-trained models and GraphControl, we align the input space across various graphs and incorporate unique characteristics of target data as conditional inputs. These conditions will be progressively integrated into the model during fine-tuning or prompt tuning through ControlNet, facilitating personalized deployment. Extensive experiments show that our method significantly enhances the adaptability of pre-trained models on target attributed datasets, achieving 1.4-3x performance gain. Furthermore, it outperforms training-from-scratch methods on target data with a comparable margin and exhibits faster convergence.
Authors: Chenguo Lin, Xumeng Wen, Wei Cao, Congrui Huang, Jiang Bian, Stephen Lin, Zhirong Wu
Recent research on time-series self-supervised models shows great promise in learning semantic representations. However, it has been limited to small-scale datasets, e.g., thousands of temporal sequences. In this work, we make key technical contributions that are tailored to the numerical properties of time-series data and allow the model to scale to large datasets, e.g., millions of temporal sequences. We adopt the Transformer architecture by first partitioning the input into non-overlapping windows. Each window is then characterized by its normalized shape and two scalar values denoting the mean and standard deviation within each window. To embed scalar values that may possess arbitrary numerical scales to high-dimensional vectors, we propose a numerically multi-scaled embedding module enumerating all possible scales for the scalar values. The model undergoes pretraining using the proposed numerically multi-scaled embedding with a simple contrastive objective on a large-scale dataset containing over a million sequences. We study its transfer performance on a number of univariate and multivariate classification benchmarks. Our method exhibits remarkable improvement against previous representation learning approaches and establishes the new state of the art, even compared with domain-specific non-learning-based methods.
Authors: Zhengmeng Xu, Hai Lin
We propose a time series forecasting method named Quantum Gramian Angular Field (QGAF). This approach merges the advantages of quantum computing technology with deep learning, aiming to enhance the precision of time series classification and forecasting. We successfully transformed stock return time series data into two-dimensional images suitable for Convolutional Neural Network (CNN) training by designing specific quantum circuits. Distinct from the classical Gramian Angular Field (GAF) approach, QGAF's uniqueness lies in eliminating the need for data normalization and inverse cosine calculations, simplifying the transformation process from time series data to two-dimensional images. To validate the effectiveness of this method, we conducted experiments on datasets from three major stock markets: the China A-share market, the Hong Kong stock market, and the US stock market. Experimental results revealed that compared to the classical GAF method, the QGAF approach significantly improved time series prediction accuracy, reducing prediction errors by an average of 25% for Mean Absolute Error (MAE) and 48% for Mean Squared Error (MSE). This research confirms the potential and promising prospects of integrating quantum computing with deep learning techniques in financial time series forecasting.
Authors: Yuyang Liu, Weijun Dong, Yingdong Hu, Chuan Wen, Zhao-Heng Yin, Chongjie Zhang, Yang Gao
Humans often acquire new skills through observation and imitation. For robotic agents, learning from the plethora of unlabeled video demonstration data available on the Internet necessitates imitating the expert without access to its action, presenting a challenge known as Imitation Learning from Observations (ILfO). A common approach to tackle ILfO problems is to convert them into inverse reinforcement learning problems, utilizing a proxy reward computed from the agent's and the expert's observations. Nonetheless, we identify that tasks characterized by a progress dependency property pose significant challenges for such approaches; in these tasks, the agent needs to initially learn the expert's preceding behaviors before mastering the subsequent ones. Our investigation reveals that the main cause is that the reward signals assigned to later steps hinder the learning of initial behaviors. To address this challenge, we present a novel ILfO framework that enables the agent to master earlier behaviors before advancing to later ones. We introduce an Automatic Discount Scheduling (ADS) mechanism that adaptively alters the discount factor in reinforcement learning during the training phase, prioritizing earlier rewards initially and gradually engaging later rewards only when the earlier behaviors have been mastered. Our experiments, conducted on nine Meta-World tasks, demonstrate that our method significantly outperforms state-of-the-art methods across all tasks, including those that are unsolvable by them.
Authors: Martin Pawelczyk, Seth Neel, Himabindu Lakkaraju
Machine unlearning, the study of efficiently removing the impact of specific training points on the trained model, has garnered increased attention of late, driven by the need to comply with privacy regulations like the Right to be Forgotten. Although unlearning is particularly relevant for LLMs in light of the copyright issues they raise, achieving precise unlearning is computationally infeasible for very large models. To this end, recent work has proposed several algorithms which approximate the removal of training data without retraining the model. These algorithms crucially rely on access to the model parameters in order to update them, an assumption that may not hold in practice due to computational constraints or when the LLM is accessed via API. In this work, we propose a new class of unlearning methods for LLMs we call ''In-Context Unlearning'', providing inputs in context and without having to update model parameters. To unlearn a particular training instance, we provide the instance alongside a flipped label and additional correctly labelled instances which are prepended as inputs to the LLM at inference time. Our experimental results demonstrate that these contexts effectively remove specific information from the training set while maintaining performance levels that are competitive with (or in some cases exceed) state-of-the-art unlearning methods that require access to the LLM parameters.
Authors: Chaoqi Liang, Weiqiang Bai, Lifeng Qiao, Yuchen Ren, Jianle Sun, Peng Ye, Hongliang Yan, Xinzhu Ma, Wangmeng Zuo, Wanli Ouyang
With the success of large-scale pretraining in NLP, there is an increasing trend of applying it to the domain of life sciences. In particular, pretraining methods based on DNA sequences have garnered growing attention due to their potential to capture generic information about genes. However, existing pretraining methods for DNA sequences largely rely on direct adoptions of BERT pretraining from NLP, lacking a comprehensive understanding and a specifically tailored approach. To address this research gap, we first conducted a series of exploratory experiments and gained several insightful observations: 1) In the fine-tuning phase of downstream tasks, when using K-mer overlapping tokenization instead of K-mer non-overlapping tokenization, both overlapping and non-overlapping pretraining weights show consistent performance improvement.2) During the pre-training process, using K-mer overlapping tokenization quickly produces clear K-mer embeddings and reduces the loss to a very low level, while using K-mer non-overlapping tokenization results in less distinct embeddings and continuously decreases the loss. 3) Using overlapping tokenization causes the self-attention in the intermediate layers of pre-trained models to tend to overly focus on certain tokens, reflecting that these layers are not adequately optimized. In summary, overlapping tokenization can benefit the fine-tuning of downstream tasks but leads to inadequate pretraining with fast convergence. To unleash the pretraining potential, we introduce a novel approach called RandomMask, which gradually increases the task difficulty of BERT-like pretraining by continuously expanding its mask boundary, forcing the model to learn more knowledge. RandomMask is simple but effective, achieving top-tier performance across 26 datasets of 28 datasets spanning 7 downstream tasks.
Authors: Minseok Choi, Hyesu Lim, Jaegul Choo
Document-level relation extraction (DocRE) aims to extract relations of all entity pairs in a document. A key challenge in DocRE is the cost of annotating such data which requires intensive human effort. Thus, we investigate the case of DocRE in a low-resource setting, and we find that existing models trained on low data overestimate the NA ("no relation") label, causing limited performance. In this work, we approach the problem from a calibration perspective and propose PRiSM, which learns to adapt logits based on relation semantic information. We evaluate our method on three DocRE datasets and demonstrate that integrating existing models with PRiSM improves performance by as much as 26.38 F1 score, while the calibration error drops as much as 36 times when trained with about 3% of data. The code is publicly available at https://github.com/brightjade/PRiSM.
Authors: Weijie Kuang, Hann Woei Ho, Ye Zhou, Shahrel Azmin Suandi, Farzad Ismail
Unmanned Aerial Vehicles (UAVs) are considered cutting-edge technology with highly cost-effective and flexible usage scenarios. Although many papers have reviewed the application of UAVs in agriculture, the review of the application for tree detection is still insufficient. This paper focuses on tree detection methods applied to UAV data collected by UAVs. There are two kinds of data, the point cloud and the images, which are acquired by the Light Detection and Ranging (LiDAR) sensor and camera, respectively. Among the detection methods using point-cloud data, this paper mainly classifies these methods according to LiDAR and Digital Aerial Photography (DAP). For the detection methods using images directly, this paper reviews these methods by whether or not to use the Deep Learning (DL) method. Our review concludes and analyses the comparison and combination between the application of LiDAR-based and DAP-based point cloud data. The performance, relative merits, and application fields of the methods are also introduced. Meanwhile, this review counts the number of tree detection studies using different methods in recent years. From our statics, the detection task using DL methods on the image has become a mainstream trend as the number of DL-based detection researches increases to 45% of the total number of tree detection studies up to 2022. As a result, this review could help and guide researchers who want to carry out tree detection on specific forests and for farmers to use UAVs in managing agriculture production.