论文检索

Deep Learning · Everything Else

Changlong Shi, Jinmeng Li, He Zhao, Dandan Guo, Yi Chang

In Federated Learning (FL), weighted aggregation of local models is conducted to generate a new global model, and the aggregation weights are typically normalized to 1. A recent study identifies the global weight shrinking effect in FL, indicating an enhancement in the global model’s generalization when the sum of weights (i.e., the shrinking factor) is smaller than 1, where how to learn the shrinking factor becomes crucial. However, principled approaches to this solution have not been carefully studied from the adequate consideration of privacy concerns and layer-wise distinctions. To this end, we propose a novel model aggregation strategy, Federated Learning with Adaptive Layer-wise Weight Shrinking (FedLWS), which adaptively designs the shrinking factor in a layer-wise manner and avoids optimizing the shrinking factors on a proxy dataset. We initially explored the factors affecting the shrinking factor during the training process. Then we calculate the layer-wise shrinking factors by considering the distinctions among each layer of the global model. FedLWS can be easily incorporated with various existing methods due to its flexibility. Extensive experiments under diverse scenarios demonstrate the superiority of our method over several state-of-the-art approaches, providing a promising tool for enhancing the global model in FL.

Theory · Deep Learning

Elvis Dohmatob, Yunzhen Feng, Arjun Subramonian, Julia Kempe

Within the scaling laws paradigm, which underpins the training of large neural networks like ChatGPT and Llama, we consider a supervised regression setting and establish a strong form of the model collapse phenomenon, a critical performance degradation due to synthetic data in the training corpus. Our results show that even the smallest fraction of synthetic data (e.g., as little as 1 per 1000) can still lead to model collapse: larger and larger training sets do not enhance performance. We further investigate whether increasing model size, an approach aligned with current trends in training large language models, exacerbates or mitigates model collapse. In a simplified regime where neural networks are approximated via random projections of tunable size, we both theoretically and empirically show that larger models can amplify model collapse. Interestingly, our theory also indicates that, beyond the interpolation threshold (which can be extremely high for very large datasets), larger models may mitigate the collapse, although they do not entirely prevent it. Our theoretical findings are empirically verified through experiments on language models and neural networks for images.

Reinforcement Learning · Online

Pouya Hamadanian, Arash Nasr-Esfahany, Malte Schwarzkopf, Siddhartha Sen, Mohammad Alizadeh

We study online reinforcement learning (RL) in non-stationary environments, where a time-varying exogenous context process affects the environment dynamics. Online RL is challenging in such environments due to "catastrophic forgetting" (CF). The agent tends to forget prior knowledge as it trains on new experiences. Prior approaches to mitigate this issue assume task labels (which are often not available in practice), employ brittle regularization heuristics, or use off-policy methods that suffer from instability and poor performance.We present Locally Constrained Policy Optimization (LCPO), an online RL approach that combats CF by anchoring policy outputs on old experiences while optimizing the return on current experiences. To perform this anchoring, LCPO locally constrains policy optimization using samples from experiences that lie outside of the current context distribution. We evaluate LCPO in Mujoco, classic control and computer systems environments with a variety of synthetic and real context traces, and find that it outperforms a variety of baselines in the non-stationary setting, while achieving results on-par with a "prescient" agent trained offline across all context traces.LCPO's source code is available at https://github.com/pouyahmdn/LCPO.

Deep Learning · Large Language Models

Chi-Heng Lin, Shangqian Gao, James Smith, Abhishek Patel, Shikhar Tuli, Yilin Shen, Hongxia Jin, Yen-Chang Hsu

Large Language Models (LLMs) have significantly advanced AI with their exceptional performance across a wide range of tasks. However, their extensive computational requirements restrict their use on devices with limited resources.While recent compression methods based on low-rank matrices show potentialsolutions, they often suffer from significant loss of accuracy or introduce substantialoverhead in parameters and inference time. In this paper, we introduce Modular De-composition (MoDeGPT), a new, efficient, and structured compression frameworkthat overcomes these limitations. MoDeGPT jointly decomposes pairs of consecu-tive subcomponents within Transformer blocks, reduces hidden dimensions throughoutput reconstruction on a larger structural scale than conventional low-rank meth-ods, and repurposes three classical matrix decomposition algorithms—Nyströmapproximation, CR decomposition, and SVD—to ensure bounded errors in ournovel decomposition approach. Our experiments show that MoDeGPT, withoutrelying on backward propagation, consistently matches or surpasses the performance of prior techniques that depend on gradient information, while achieving a98% reduction in compute costs when compressing a 13B-parameter model. OnLLaMA-2/3 and OPT models, MoDeGPT retains 90-95% of zero-shot performancewith compression rates of 25-30%. The compression process can be completed ona single GPU in a few hours, boosting inference throughput by up to 46%.

Deep Learning · Large Language Models

Chi-Heng Lin, Shangqian Gao, James Smith, Abhishek Patel, Shikhar Tuli, Yilin Shen, Hongxia Jin, Yen-Chang Hsu

Large Language Models (LLMs) have significantly advanced AI with their exceptional performance across a wide range of tasks. However, their extensive computational requirements restrict their use on devices with limited resources.While recent compression methods based on low-rank matrices show potentialsolutions, they often suffer from significant loss of accuracy or introduce substantialoverhead in parameters and inference time. In this paper, we introduce Modular De-composition (MoDeGPT), a new, efficient, and structured compression frameworkthat overcomes these limitations. MoDeGPT jointly decomposes pairs of consecu-tive subcomponents within Transformer blocks, reduces hidden dimensions throughoutput reconstruction on a larger structural scale than conventional low-rank meth-ods, and repurposes three classical matrix decomposition algorithms—Nyströmapproximation, CR decomposition, and SVD—to ensure bounded errors in ournovel decomposition approach. Our experiments show that MoDeGPT, withoutrelying on backward propagation, consistently matches or surpasses the performance of prior techniques that depend on gradient information, while achieving a98% reduction in compute costs when compressing a 13B-parameter model. OnLLaMA-2/3 and OPT models, MoDeGPT retains 90-95% of zero-shot performancewith compression rates of 25-30%. The compression process can be completed ona single GPU in a few hours, boosting inference throughput by up to 46%.

Probabilistic Methods · Bayesian Models and Methods

João Loula, Benjamin LeBrun, Li Du, Ben Lipkin, Clemente Pasti, Gabriel Grand, Tianyu Liu, Yahya Emara, Marjorie Freedman, Jason Eisner 等

A wide range of LM applications require generating text that conforms to syntactic or semantic constraints. Imposing such constraints can be naturally framed as _probabilistic conditioning_, but exact generation from the resulting distribution—which can differ substantially from the LM’s base distribution—is generally intractable. In this work,we develop an architecture for controlled LM generation based on sequential Monte Carlo (SMC). Our SMC framework allows us to flexibly incorporate domain- and problem-specific constraints at inference time, and efficiently reallocate computational resources in light of new information during the course of generation. By comparing to a number of alternatives and ablations on four challenging domains---Python code generation for data science, text-to-SQL, goal inference, and molecule synthesis—we demonstrate that, with little overhead, our approach allows small open-source language models to outperform models over 8$\times$ larger, as well as closed-source, fine-tuned ones. In support of the probabilistic perspective, we show that these performance improvements are driven by better approximation to the posterior distribution. [Our system](https://github.com/probcomp/genlm-control) builds on the framework of Lew et al. (2023) and integrates with its _language model probabilistic programming language_, giving users a simple, programmable way to apply SMC to a broad variety of controlled generation problems.

João Loula, Benjamin LeBrun, Li Du, Ben Lipkin, Clemente Pasti, Gabriel Grand, Tianyu Liu, Yahya Emara, Marjorie Freedman, Jason Eisner 等

A wide range of LM applications require generating text that conforms to syntactic or semantic constraints. Imposing such constraints can be naturally framed as probabilistic conditioning, but exact generation from the resulting distribution—which can differ substantially from the LM’s base distribution—is generally intractable. In this work, we develop an architecture for controlled LM generation based on sequential Monte Carlo (SMC). This SMC framework allows us to flexibly incorporate domain- and problem-specific constraints at inferencetime, and efficiently reallocate computational resources in light of new information during the course of generation. By comparing to a number of alternatives and ablations on four challenging domains—Python code generation for data science, text-to-SQL, goal inference, and molecule synthesis—we demonstrate that, with little overhead, our approach allows small open-source language models to outperform models over 8× larger, as well as closed-source, fine-tuned ones. In support of the probabilistic perspective, we show that these performance improvements are driven by better approximation to the posterior distribution. [Our system](https://github.com/probcomp/gen-parse) builds on the framework of Lew et al. (2023) and integrates with its language model probabilistic programming language, giving users a simple, programmable way to apply SMC to a broad variety of controlled generation problems.

Reinforcement Learning · Inverse

Bo Yue, Shufan Wang, Ashish Gaurav, Jian Li, Pascal Poupart, Guiliang Liu

In practical applications, the underlying constraint knowledge is often unknown and difficult to specify. To address this issue, recent advances in Inverse Constrained Reinforcement Learning (ICRL) have focused on inferring these constraints from expert demonstrations. However, the ICRL approach typically characterizes constraint learning as a tri-level optimization problem, which is inherently complex due to its interdependent variables and multiple layers of optimization.Considering these challenges, a critical question arises: *Can we implicitly embed constraint signals into reward functions and effectively solve this problem using a classic reward inference algorithm?* The resulting method, known as Inverse Reward Correction (IRC), merits investigation. In this work, we conduct a theoretical analysis comparing the sample complexities of both solvers. Our findings confirm that the IRC solver achieves lower sample complexity than its ICRL counterpart.Nevertheless, this reduction in complexity comes at the expense of generalizability. Specifically, in the target environment, the reward correction terms may fail to guarantee the safety of the resulting policy, whereas this issue can be effectively mitigated by transferring the constraints via the ICRL solver. Advancing our inquiry, we investigate conditions under which the ICRL solver ensures $\epsilon$-optimality when transferring to new environments. Empirical results across various environments validate our theoretical findings, underscoring the nuanced trade-offs between complexity reduction and generalizability in safety-critical applications.

Theory · Learning Theory

Yuchen Liang, Peizhong Ju, Yingbin Liang, Ness Shroff

Accelerated diffusion models hold the potential to significantly enhance the efficiency of standard diffusion processes. Theoretically, these models have been shown to achieve faster convergence rates than the standard $\mathcal O(1/\epsilon^2)$ rate of vanilla diffusion models, where $\epsilon$ denotes the target accuracy. However, current theoretical studies have established the acceleration advantage only for restrictive target distribution classes, such as those with smoothness conditions imposed along the entire sampling path or with bounded support. In this work, we significantly broaden the target distribution classes with a new accelerated stochastic DDPM sampler. In particular, we show that it achieves accelerated performance for three broad distribution classes not considered before. Our first class relies on the smoothness condition posed only to the target density $q_0$, which is far more relaxed than the existing smoothness conditions posed to all $q_t$ along the entire sampling path. Our second class requires only a finite second moment condition, allowing for a much wider class of target distributions than the existing finite-support condition. Our third class is Gaussian mixture, for which our result establishes the first acceleration guarantee. Moreover, among accelerated DDPM type samplers, our results specialized for bounded-support distributions show an improved dependency on the data dimension $d$. Our analysis introduces a novel technique for establishing performance guarantees via constructing a tilting factor representation of the convergence error and utilizing Tweedie's formula to handle Taylor expansion terms. This new analytical framework may be of independent interest.

Applications · Computer Vision

Shoubin Yu, Jaehong Yoon, Mohit Bansal

Despite impressive advancements in recent multimodal reasoning approaches, they are still limited in flexibility and efficiency, as these models typically process only a few fixed modality inputs and require updates to numerous parameters. This paper tackles these critical challenges and proposes CREMA, a generalizable, highly efficient, and modular modality-fusion framework that can incorporate many new modalities to enhance video reasoning. We first augment multiple informative modalities (such as optical flow, 3D point cloud, audio, thermal heatmap, and touch map) from given videos without extra human annotation by leveraging sensors or existing pre-trained models. Next, we introduce a query transformer with multiple parameter-efficient modules associated with each accessible modality. It projects diverse modality features to the LLM token embedding space, allowing the model to integrate different data types for response generation. Furthermore, we propose a novel progressive multimodal fusion design supported by a lightweight fusion module and modality-sequential training strategy. It helps compress information across various assisting modalities, maintaining computational efficiency in the LLM while improving performance. We validate our method on seven video-language reasoning tasks assisted by diverse modalities, including conventional VideoQA and Video-Audio/3D/Touch/Thermal QA, and achieve better/equivalent performance against strong multimodal LLMs, including OneLLM, BLIP-2, and SeViLA while reducing over 90% trainable parameters. We provide extensive analyses of CREMA, including the impact of each modality on reasoning domains, the design of the fusion module, and example visualizations.

Applications · Computer Vision

Shangzhe Di, Zhelun Yu, Guanghao Zhang, Haoyuan Li, ZhongTao, Hao Cheng, Bolin Li, Wanggui He, Fangxun Shu, Hao Jiang

We propose ReKV, a novel training-free approach that enables efficient streaming video question-answering (StreamingVQA), by seamlessly integrating with existing Video Large Language Models (Video-LLMs). Traditional VideoQA systems struggle with long videos, as they must process entire videos before responding to queries, and repeat this process for each new question. In contrast, our approach analyzes long videos in a streaming manner, allowing for prompt responses as soon as user queries are received. Building on a common Video-LLM, we first incorporate a sliding-window attention mechanism, ensuring that input frames attend to a limited number of preceding frames, thereby reducing computational overhead. To prevent information loss, we store processed video key-value caches (KV-Caches) in RAM and disk, reloading them into GPU memory as needed. Additionally, we introduce a retrieval method that leverages an external retriever or the parameters within Video-LLMs to retrieve only query-relevant KV-Caches, ensuring both efficiency and accuracy in question answering. ReKV enables the separation of video analyzing and question-answering across different processes and GPUs, significantly enhancing the efficiency of StreamingVQA. Through comprehensive experimentation, we validate the efficacy and practicality of our approach, which significantly boosts efficiency and enhances applicability over existing VideoQA models.

Reinforcement Learning · Deep RL

Giseung Park, Youngchul Sung

In this paper, we introduce a simple yet effective reward dimension reduction method to tackle the scalability challenges of multi-objective reinforcement learning algorithms. While most existing approaches focus on optimizing two to four objectives, their abilities to scale to environments with more objectives remain uncertain. Our method uses a dimension reduction approach to enhance learning efficiency and policy performance in multi-objective settings. While most traditional dimension reduction methods are designed for static datasets, our approach is tailored for online learning and preserves Pareto-optimality after transformation. We propose a new training and evaluation framework for reward dimension reduction in multi-objective reinforcement learning and demonstrate the superiority of our method in environments including one with sixteen objectives, significantly outperforming existing online dimension reduction methods.

Applications · Computer Vision

Rachel Mikulinsky, Morris Alper, Shai Gordin, Enrique Jiménez, Yoram Cohen, Hadar Averbuch-Elor

The cuneiform writing system served as the medium for transmitting knowledgein the ancient Near East for a period of over three thousand years. Cuneiformsigns have a complex internal structure which is the subject of expert paleographicanalysis, as variations in sign shapes bear witness to historical developments andtransmission of writing and culture over time. However, prior automated techniquesmostly treat sign types as categorical and do not explicitly model their highly variedinternal configurations. In this work, we present an unsupervised approach forrecovering the fine-grained internal configuration of cuneiform signs by leveragingpowerful generative models and the appearance and structure of prototype fontimages as priors. Our approach, ProtoSnap, enforces structural consistency onmatches found with deep image features to estimate the diverse configurationsof cuneiform characters, snapping a skeleton-based template to photographedcuneiform signs. We provide a new benchmark of expert annotations and evaluateour method on this task. Our evaluation shows that our approach succeeds inaligning prototype skeletons to a wide variety of cuneiform signs. Moreover, weshow that conditioning on structures produced by our method allows for generatingsynthetic data with correct structural configurations, significantly boosting theperformance of cuneiform sign recognition beyond existing techniques, in particularover rare signs. Our code, data, and trained models are available at the project page:https://tau-vailab.github.io/ProtoSnap/

Reinforcement Learning · Inverse

Runzhe Wu, Yiding Chen, Gokul Swamy, Kianté Brantley, Wen Sun

Adversarial Imitation Learning is traditionally framed as a two-player zero-sum game between a learner and an adversarially chosen cost function, and can therefore be thought of as the sequential generalization of a Generative Adversarial Network (GAN). However, in recent years, diffusion models have emerged as a non-adversarial alternative to GANs that merely require training a score function via regression, yet produce generations of higher quality. In response, we investigate how to lift insights from diffusion modeling to the sequential setting. We propose diffusing states and performing *score-matching* along diffused states to measure the discrepancy between the expert's and learner's states. Thus, our approach only requires training score functions to predict noises via standard regression, making it significantly easier and more stable to train than adversarial methods. Theoretically, we prove first- and second-order instance-dependent bounds with linear scaling in the horizon, proving that our approach avoids the compounding errors that stymie offline approaches to imitation learning. Empirically, we show our approach outperforms both GAN-style imitation learning baselines and discriminator-free imitation learning baselines across various continuous control problems, including complex tasks like controlling humanoids to walk, sit, crawl, and navigate through obstacles.

Applications · Computer Vision

Shivam Duggal, Phillip Isola, Antonio Torralba, William Freeman

Current vision systems typically assign fixed-length representations to images, regardless of the information content. This contrasts with human intelligence —and even large language models—which allocate varying representational capacities based on entropy, context and familiarity. Inspired by this, we propose an approach to learn variable-length token representations for 2D images. Our encoder-decoder architecture recursively processes 2D image tokens, distilling them into 1D latent tokens over multiple iterations of recurrent rollouts. Each iteration refines the 2D tokens, updates the existing 1D latent tokens, and adaptively increases representational capacity by adding new tokens. This enables compression of images into a variable number of tokens, ranging from 32 to 256. We validate our tokenizer using reconstruction loss and FID metrics, demonstrating that token count aligns with image entropy, familiarity and downstream task requirements. Recurrent token processing with increasing representational capacity in each iteration shows signs of token specialization, revealing potential for object / part discovery.

Deep Learning · Generative Models and Autoencoders

Ryien Hosseini, Filippo Simini, Venkatram Vishwanath, Rebecca Willett, Henry Hoffmann

Deep generative models have recently achieved significant success in modeling graph data, including dynamic graphs, where topology and features evolve over time. However, unlike in vision and natural language domains, evaluating generative models for dynamic graphs is challenging due to the difficulty of visualizing their output, making quantitative metrics essential. In this work, we develop a new quality metric for evaluating generative models of dynamic graphs. Current metrics for dynamic graphs typically involve discretizing the continuous-evolution of graphs into static snapshots and then applying conventional graph similarity measures. This approach has several limitations: (a) it models temporally related events as i.i.d. samples, failing to capture the non-uniform evolution of dynamic graphs; (b) it lacks a unified measure that is sensitive to both features and topology; (c) it fails to provide a scalar metric, requiring multiple metrics without clear superiority; and (d) it requires explicitly instantiating each static snapshot, leading to impractical runtime demands that hinder evaluation at scale. We propose a novel metric based on the Johnson-Lindenstrauss lemma, applying random projections directly to dynamic graph data. This results in an expressive, scalar, and application-agnostic measure of dynamic graph similarity that overcomes the limitations of traditional methods. We also provide a comprehensive empirical evaluation of metrics for continuous-time dynamic graphs, demonstrating the effectiveness of our approach compared to existing methods. Our implementation is available at https://github.com/ryienh/jl-metric.

Optimization · Sampling and Optimization

Sebastian Sanokowski, Wilhelm Berghammer, Haoyu Wang, Martin Ennemoser, Sepp Hochreiter, Sebastian Lehner

Learning to sample from complex unnormalized distributions over discrete domains emerged as a promising research direction with applications in statistical physics, variational inference, and combinatorial optimization. Recent work has demonstrated the potential of diffusion models in this domain. However, existing methods face limitations in memory scaling and thus the number of attainable diffusion steps since they require backpropagation through the entire generative process. To overcome these limitations we introduce two novel training methods for discrete diffusion samplers, one grounded in the policy gradient theorem and the other one leveraging Self-Normalized Neural Importance Sampling (SN-NIS). These methods yield memory-efficient training and achieve state-of-the-art results in unsupervised combinatorial optimization.Numerous scientific applications additionally require the ability of unbiased sampling. We introduce adaptations of SN-NIS and Neural Markov Chain Monte Carlo that enable for the first time the application of discrete diffusion models to this problem. We validate our methods on Ising model benchmarks and find that they outperform popular autoregressive approaches. Our work opens new avenues for applying diffusion models to a wide range of scientific applications in discrete domains that were hitherto restricted to exact likelihood models.

Theory · Online Learning and Bandits

Zhuohua Li, Maoli Liu, Xiangxiang Dai, John C.S. Lui

The contextual multi-armed bandit (MAB) problem is crucial in sequential decision-making. A line of research, known as online clustering of bandits, extends contextual MAB by grouping similar users into clusters, utilizing shared features to improve learning efficiency. However, existing algorithms, which rely on the upper confidence bound (UCB) strategy, struggle to gather adequate statistical information to accurately identify unknown user clusters. As a result, their theoretical analyses require several strong assumptions about the "diversity" of contexts generated by the environment, leading to impractical settings, complicated analyses, and poor practical performance. Removing these assumptions has been a long-standing open problem in the clustering of bandits literature. In this work, we provide two partial solutions. First, we introduce an additional exploration phase to accelerate the identification of clusters. We integrate this general strategy into both graph-based and set-based algorithms and propose two new algorithms, UniCLUB and UniSCLUB. Remarkably, our algorithms require substantially weaker assumptions and simpler theoretical analyses while achieving superior cumulative regret compared to previous studies. Second, inspired by the smoothed analysis framework, we propose a more practical setting that eliminates the requirement for i.i.d. context generation used in previous studies, thus enhancing the performance of existing algorithms for online clustering of bandits. Extensive evaluations on both synthetic and real-world datasets demonstrate that our proposed algorithms outperform existing approaches.

Applications · Robotics

Patrick Yin, Tyler Westenbroek, Ching-An Cheng, Andrey Kolobov, Abhishek Gupta

Robot learning requires a considerable amount of high-quality data to realize the promise of generalization. However, large data sets are costly to collect in the real world. Physics simulators can cheaply generate vast data sets with broad coverage over states, actions, and environments. However, physics engines are fundamentally misspecified approximations to reality. This makes direct zero-shot transfer from simulation to reality challenging, especially in tasks where precise and force-sensitive manipulation is necessary. Thus, fine-tuning these policies with small real-world data sets is an appealing pathway for scaling robot learning. However, current reinforcement learning fine-tuning frameworks leverage general, unstructured exploration strategies which are too inefficient to make real-world adaptation practical. This paper introduces the \emph{Simulation-Guided Fine-tuning} (SGFT) framework, which demonstrates how to extract structural priors from physics simulators to substantially accelerate real-world adaptation. Specifically, our approach uses a value function learned in simulation to guide real-world exploration. We demonstrate this approach across five real-world dexterous manipulation tasks where zero-shot sim-to-real transfer fails. We further demonstrate our framework substantially outperforms baseline fine-tuning methods, requiring up to an order of magnitude fewer real-world samples and succeeding at difficult tasks where prior approaches fail entirely. Last but not least, we provide theoretical justification for this new paradigm which underpins how SGFT can rapidly learn high-performance policies in the face of large sim-to-real dynamics gaps.

Deep Learning · Sequential Models, Time series

Ying-yee Ava Lau, Zhiwen Shao, Dit-Yan Yeung

Current research in online time series forecasting (OTSF) faces two significant issues. The first is information leakage, where models make predictions and are then evaluated on historical time steps that have already been used in backpropagation for parameter updates. The second is practicality: while forecasting in real-world applications typically emphasizes looking ahead and anticipating future uncertainties, prediction sequences in this setting include only one future step with the remaining being observed time points. This necessitates a redefinition of the OTSF setting, focusing on predicting unknown future steps and evaluating unobserved data points. Following this new setting, challenges arise in leveraging incomplete pairs of ground truth and predictions for backpropagation, as well as in generalizing accurate information without overfitting to noise from recent data streams. To address these challenges, we propose a novel dual-stream framework for online forecasting (DSOF): a slow stream that updates with complete data using experience replay, and a fast stream that adapts to recent data through temporal difference learning. This dual-stream approach updates a teacher-student model learned through a residual learning strategy, generating predictions in a coarse-to-fine manner. Extensive experiments demonstrate its improvement in forecasting performance in changing environments. Our code is publicly available at https://github.com/yyalau/iclr2025_dsof.