ICLR 2025 论文列表

Developing reinforcement learning agents that can generalise effectively to new tasks is one of the main challenges in AI research. This paper introduces Fracture Cluster Options (FraCOs), a multi-level hierarchical reinforcement learning method designed to improve generalisation performance. FraCOs identifies patterns in agent behaviour and forms temporally-extended actions (options) based on the expected future usefulness of those patterns, enabling rapid adaptation to new tasks. In tabular settings, FraCOs demonstrates effective transfer and improves performance as the depth of the hierarchy increases. In several complex procedurally-generated environments, FraCOs consistently outperforms state-of-the-art deep reinforcement learning algorithms, achieving superior results in both in-distribution and out-of-distribution scenarios.

Echo State Networks (ESNs) are a class of recurrent neural networks in which only a small readout regression layer is trained, while the weights of the recurrent network, termed the reservoir, are randomly assigned and remain fixed. Our work introduces the Locally Connected ESN (LCESN), a novel ESN variant with a locally connected reservoir, forced memory, and a weight adaptation strategy. LCESN significantly reduces the asymptotic time and space complexities compared to the conventional ESN, enabling substantially larger networks. LCESN also improves the memory properties of ESNs without affecting network stability. We evaluate LCESN's performance on the NARMA10 benchmark task and compare it to state-of-the-art models on nine real-world datasets. Despite the simplicity of our model and its one-shot training approach, LCESN achieves competitive results, even surpassing several state-of-the-art models. LCESN introduces a fresh approach to real-world time series forecasting and demonstrates that large, well-tuned random recurrent networks can rival complex gradient-trained feedforward models. We provide our GPU-based implementation of LCESN as an open-source library.

Contrastive Learning (CL) proves to be effective for learning generalizable user representations in Sequential Recommendation (SR), but it suffers from high computational costs due to its reliance on negative samples. To overcome this limitation, we propose the first Non-Contrastive Learning (NCL) framework for SR, which eliminates computational overhead of identifying and generating negative samples. However, without negative samples, it is challenging to learn uniform representations from only positive samples, which is prone to representation collapse. Furthermore, the alignment of the learned representations may be substantially compromised because existing ad-hoc augmentations can produce positive samples that have inconsistent user preferences. To tackle these challenges, we design a novel preference-preserving profile generation method to produce high-quality positive samples for non-contrastive training. Inspired by differential privacy, our approach creates augmented user profiles that exhibit high diversity while provably retaining consistent user preferences. With larger diversity and consistency of the positive samples, our NCL framework significantly enhances the alignment and uniformity of the learned representations, which contributes to better generalization. The experimental results on various benchmark datasets and model architectures demonstrate the effectiveness of the proposed method. Finally, our investigations reveal that both uniformity and alignment play a vital role in improving generalization for SR. Interestingly, in our data-sparse setting, alignment is usually more important than uniformity.

Recent advances in vision-language models (VLMs) have made significant progress in downstream tasks that require quantitative concepts such as facial age estimation and image quality assessment, enabling VLMs to explore applications like image ranking and retrieval. However, existing studies typically focus on the reasoning based on a single image and heavily depend on text prompting, limiting their ability to learn comprehensive understanding from multiple images. To address this, we propose an effective yet efficient approach that reframes the CLIP model into a learning-to-rank task and introduces a lightweight adapter to augment CLIP for text-guided image ranking. Specifically, our approach incorporates learnable prompts to adapt to new instructions for ranking purposes and an auxiliary branch with ranking-aware attention, leveraging text-conditioned visual differences for additional supervision in image ranking. Our ranking-aware adapter consistently outperforms fine-tuned CLIPs on various tasks and achieves competitive results compared to state-of-the-art models designed for specific tasks like facial age estimation and image quality assessment. Overall, our approach primarily focuses on ranking images with a single instruction, which provides a natural and generalized way of learning from visual differences across images, bypassing the need for extensive text prompts tailored to individual tasks.

Despite tremendous progress in the field of text-to-video (T2V) synthesis, open-sourced T2V diffusion models struggle to generate longer videos with dynamically varying and evolving content. They tend to synthesize quasi-static videos, ignoring the necessary visual change-over-time implied in the text prompt. Meanwhile, scaling these models to enable longer, more dynamic video synthesis often remains computationally intractable. To tackle this challenge, we introduce the concept of Generative Temporal Nursing (GTN), where we adjust the generative process on the fly during inference to improve control over the temporal dynamics and enable generation of longer videos. We propose a method for GTN, dubbed VSTAR, which consists of two key ingredients: Video Synopsis Prompting (VSP) and Temporal Attention Regularization (TAR), the latter being our core contribution. Based on a systematic analysis, we discover that the temporal units in pretrained T2V models are crucial to control the video dynamics. Upon this finding, we propose a novel regularization technique to refine the temporal attention, enabling training-free longer video synthesis in a single inference pass. For prompts involving visual progression, we leverage LLMs to generate video synopsis - description of key visual states - based on the original prompt to provide better guidance along the temporal axis. We experimentally showcase the superiority of our method in synthesizing longer, visually appealing videos over open-sourced T2V models.

Large Language Models (LLMs) have achieved state-of-the-art performance across numerous tasks. However, these advancements have predominantly benefited "first-class" languages such as English and Chinese, leaving many other languages underrepresented. This imbalance, while limiting broader applications, generates a natural preference ranking between languages, offering an opportunity to bootstrap the multilingual capabilities of LLM in a self-improving manner. Thus, we propose $\textit{Language Imbalance Driven Rewarding}$, where the inherent imbalance between dominant and non-dominant languages within LLMs is leveraged as a reward signal. Iterative DPO training demonstrates that this approach not only enhances LLM performance in non-dominant languages but also improves the dominant language's capacity, thereby yielding an iterative reward signal. Fine-tuning Meta-Llama-3-8B-Instruct over two iterations of this approach results in continuous improvements in multilingual performance across instruction-following and arithmetic reasoning tasks, evidenced by an average improvement of 7.46\% win rate on the X-AlpacaEval leaderboard and 13.9\% accuracy on the MGSM benchmark. This work serves as an initial exploration, paving the way for multilingual self-improvement of LLMs.

We introduce a novel class of regularization functions, called Cauchy–Schwarz (CS) regularizers, which can be designed to induce a wide range of properties in solution vectors of optimization problems. To demonstrate the versatility of CS regularizers, we derive regularization functions that promote discrete-valued vectors, eigenvectors of a given matrix, and orthogonal matrices. The resulting CS regularizers are simple, differentiable, and can be free of spurious stationary points, making them suitable for gradient-based solvers and large-scale optimization problems. In addition, CS regularizers automatically adapt to the appropriate scale, which is, for example, beneficial when discretizing the weights of neural networks. To demonstrate the efficacy of CS regularizers, we provide results for solving underdetermined systems of linear equations and weight quantization in neural networks. Furthermore, we discuss specializations, variations, and generalizations, which lead to an even broader class of new and possibly more powerful regularizers.

Large Language Models (LLMs) have become instrumental across various applications, with the customization of these models to specific scenarios becoming increasingly critical. System message, a fundamental component of LLMs, is consist of carefully crafted instructions that guide the behavior of model to meet intended goals. Despite the recognized potential of system messages to optimize AI-driven solutions, there is a notable absence of a comprehensive benchmark for evaluating how well LLMs follow system messages. To fill this gap, we introduce SysBench, a benchmark that systematically analyzes system message following ability in terms of three limitations of existing LLMs: constraint violation, instruction misjudgement and multi-turn instability. Specifically, we manually construct evaluation dataset based on six prevalent types of constraints, including 500 tailor-designed system messages and multi-turn user conversations covering various interaction relationships. Additionally, we develop a comprehensive evaluation protocol to measure model performance. Finally, we conduct extensive evaluation across various existing LLMs, measuring their ability to follow specified constraints given in system messages. The results highlight both the strengths and weaknesses of existing models, offering key insights and directions for future research.

We study "selective" or "conditional" classification problems under an agnostic setting. Classification tasks commonly focus on modeling the relationship between features and categories that captures the vast majority of data. In contrast to common machine learning frameworks, conditional classification intends to model such relationships only on a subset of the data defined by some selection rule. Most work on conditional classification either solves the problem in a realizable setting or does not guarantee the error is bounded compared to an optimal solution. In this work, we consider selective/conditional classification by sparse linear classifiers for subsets defined by halfspaces, and give both positive as well as negative results for Gaussian feature distributions. On the positive side, we present the first PAC-learning algorithm for homogeneous halfspace selectors with error guarantee $\tilde{O}(\sqrt{\mathrm{opt}})$, where $\mathrm{opt}$ is the smallest conditional classification error over the given class of classifiers and homogeneous halfspaces. On the negative side, we find that, under cryptographic assumptions, approximating the conditional classification loss within a small additive error is computationally hard even under Gaussian distribution. We prove that approximating conditional classification is at least as hard as approximating agnostic classification in both additive and multiplicative form.

We propose GALA, a novel representation of 3D shapes that (i) excels at capturing and reproducing complex geometry and surface details, (ii) is computationally efficient, and (iii) lends itself to 3D generative modelling with modern, diffusion-based schemes. The key idea of GALA is to exploit both the global sparsity of surfaces within a 3D volume and their local surface properties. *Sparsity* is promoted by covering only the 3D object boundaries, not empty space, with an ensemble of tree root voxels. Each voxel contains an octree to further limit storage and compute to regions that contain surfaces. *Adaptivity* is achieved by fitting one local and geometry-aware coordinate frame in each non-empty leaf node. Adjusting the orientation of the local grid, as well as the anisotropic scales of its axes, to the local surface shape greatly increases the amount of detail that can be stored in a given amount of memory, which in turn allows for quantization without loss of quality. With our optimized C++/CUDA implementation, GALA can be fitted to an object in less than 10 seconds. Moreover, the representation can efficiently be flattened and manipulated with transformer networks. We provide a cascaded generation pipeline capable of generating 3D shapes with great geometric detail. For more information, please visit our [project page](https://santisy.github.io/GALA/).

Protein language (or sequence) models, like the popular ESM2, are now widely used tools for extracting evolution-based protein representations and have achieved significant success on core downstream biological tasks. A major open problem is how to obtain representations that best capture both the sequence evolutionary history and the atomic structural properties of proteins in general. We introduce **I**mplicit **S**equence **M**odel, a sequence-only input model with structurally-enriched representations that outperforms state-of-the-art sequence models on several well-studied benchmarks including mutation stability assessment and structure prediction. Our key innovations are a microenvironment-based Autoencoder for generating structure tokens and a self-supervised training objective that distills these tokens into ESM2's pre-trained model. Notably, we make ISM's structure-enriched weights easily accessible for any application using the ESM2 framework.

Identifying and localizing objects within images is a fundamental challenge, and numerous efforts have been made to enhance model accuracy by experimenting with diverse architectures and refining training strategies. Nevertheless, a prevalent limitation in existing models is overemphasizing the current input while ignoring the information from the entire dataset. We introduce an innovative $\textbf{R}etriever-\textbf{D}ictionary$ (RD) module to address this issue. This architecture enables YOLO-based models to efficiently retrieve features from a Dictionary that contains the insight of the dataset, which is built by the knowledge from Visual Models (VM), Large Language Models (LLM), or Visual Language Models (VLM). The flexible RD enables the model to incorporate such explicit knowledge that enhances the ability to benefit multiple tasks, specifically, segmentation, detection, and classification, from pixel to image level. The experiments show that using the RD significantly improves model performance, achieving more than a 3\% increase in mean Average Precision for object detection with less than a 1\% increase in model parameters. Beyond 1-stage object detection models, the RD module improves the effectiveness of 2-stage models and DETR-based architectures, such as Faster R-CNN and Deformable DETR. Code is released at https://github.com/henrytsui000/YOLO.

Optimizing computing circuits such as multipliers and adders is a fundamental challenge in modern integrated circuit design. Recent efforts propose formulating this optimization problem as a reinforcement learning (RL) proxy task, offering a promising approach to search high-speed and area-efficient circuit design solutions. However, we show that the RL-based formulation (proxy task) converges to a local optimal design solution (original task) due to the deceptive reward signals and incrementally localized actions in the RL-based formulation. To address this challenge, we propose a novel model-based circuit genetic evolution (MUTE) framework, which reformulates the problem as a genetic evolution process by proposing a grid-based genetic representation of design solutions. This novel formulation avoids misleading rewards by evaluating and improving generated solutions using the true objective value rather than proxy rewards. To promote globally diverse exploration, MUTE proposes a multi-granularity genetic crossover operator that recombines design substructures at varying column ranges between two grid-based genetic solutions. To the best of our knowledge, MUTE is the first to reformulate the problem as a circuit genetic evolution process, which enables effectively searching for global optimal design solutions. We evaluate MUTE on several fundamental computing circuits, including multipliers, adders, and multiply-accumulate circuits. Experiments on these circuits demonstrate that MUTE significantly Pareto-dominates state-of-the-art approaches in terms of both area and delay. Moreover, experiments demonstrate that circuits designed by MUTE well generalize to large-scale computation-intensive circuits as well.

Ensemble methods are commonly used for enhancing robustness in machine learning. However, due to the ''transferability'' of adversarial examples, the performance of an ensemble model can be seriously affected even it contains a set of independently trained sub-models. To address this issue, we propose an efficient data transformation method based on a cute ''weakness allocation'' strategy, to diversify non-robust features. Our approach relies on a fine-grained analysis on the relation between non-robust features and adversarial attack directions. Moreover, our approach enjoys several other advantages, e.g., it does not require any communication between sub-models and the construction complexity is also quite low. We conduct a set of experiments to evaluate the performance of our proposed method and compare it with several popular baselines. The results suggest that our approach can achieve significantly improved robust accuracy over most existing ensemble methods, and meanwhile preserve high clean accuracy.

Segmenting an object in a video presents significant challenges. Each pixel must be accurately labeled, and these labels must remain consistent across frames. The difficulty increases when the segmentation is with arbitrary granularity, meaning the number of segments can vary arbitrarily, and masks are defined based on only one or a few sample images. In this paper, we address this issue by employing a pre-trained text to image diffusion model supplemented with an additional tracking mechanism. We demonstrate that our approach can effectively manage various segmentation scenarios and outperforms state-of-the-art alternatives. The project page is available at https://segment-me-in-time.github.io/

Advanced sensors and IoT devices have improved the monitoring and control of complex industrial enterprises. They have also created an interdependent fabric of geographically distributed process operations (clients) across these enterprises. Granger causality is an effective approach to detect and quantify interdependencies by examining how the state of one client affects the states of others over time. Understanding these interdependencies helps capture how localized events, such as faults and disruptions, can propagate throughout the system, potentially leading to widespread operational impacts. However, the large volume and complexity of industrial data present significant challenges in effectively modeling these interdependencies. This paper develops a federated approach to learning Granger causality. We utilize a linear state space system framework that leverages low-dimensional state estimates to analyze interdependencies. This helps address bandwidth limitations and the computational burden commonly associated with centralized data processing. We propose augmenting the client models with the Granger causality information learned by the server through a Machine Learning (ML) function. We examine the co-dependence between the augmented client and server models and reformulate the framework as a standalone ML algorithm providing conditions for its sublinear and linear convergence rates. We also study the convergence of the framework to a centralized oracle model. Moreover, we include a differential privacy analysis to ensure data security while preserving causal insights. Using synthetic data, we conduct comprehensive experiments to demonstrate the robustness of our approach to perturbations in causality, the scalability to the size of communication, number of clients, and the dimensions of raw data. We also evaluate the performance on two real-world industrial control system datasets by reporting the volume of data saved by decentralization.

Frontier AI systems are making transformative impacts across society, but such benefits are not without costs: models trained on web-scale datasets containing personal and private data raise profound concerns about data privacy and security. Language models are trained on extensive corpora including potentially sensitive or proprietary information, and the risk of data leakage, where the model response reveals pieces of such information, remains inadequately understood. Prior work has investigated that sequence complexity and the number of repetitions are the primary drivers of memorization. In this work, we examine the most vulnerable class of data: highly complex sequences that are presented only once during training. These sequences often contain the most sensitive information and pose considerable risk if memorized. By analyzing the progression of memorization for these sequences throughout training, we uncover a striking observation: many memorized sequences persist in the model's memory, exhibiting resistance to catastrophic forgetting even after just one encounter. Surprisingly, these sequences may not appear memorized immediately after their first exposure but can later be “uncovered” during training, even in the absence of subsequent exposures - a phenomenon we call "latent memorization." Latent memorization presents a serious challenge for data privacy, as sequences that seem hidden at the final checkpoint of a model may still be easily recoverable. We demonstrate how these hidden sequences can be revealed through random weight perturbations, and we introduce a diagnostic test based on cross-entropy loss to accurately identify latent memorized sequences.

In this work, we address the cooperation problem among large language model (LLM) based embodied agents, where agents must cooperate to achieve a common goal. Previous methods often execute actions extemporaneously and incoherently, without long-term strategic and cooperative planning, leading to redundant steps, failures, and even serious repercussions in complex tasks like search-and-rescue missions where discussion and cooperative plan are crucial. To solve this issue, we propose Cooperative Plan Optimization (CaPo) to enhance the cooperation efficiency of LLM-based embodied agents. Inspired by human cooperation schemes, CaPo improves cooperation efficiency with two phases: 1) meta plan generation, and 2) progress-adaptive meta plan and execution. In the first phase, all agents analyze the task, discuss, and cooperatively create a meta-plan that decomposes the task into subtasks with detailed steps, ensuring a long-term strategic and coherent plan for efficient coordination. In the second phase, agents execute tasks according to the meta-plan and dynamically adjust it based on their latest progress (e.g., discovering a target object) through multi-turn discussions. This progress-based adaptation eliminates redundant actions, improving the overall cooperation efficiency of agents. Experimental results on the ThreeDworld Multi-Agent Transport and Communicative Watch-And-Help tasks demonstrate CaPo's much higher task completion rate and efficiency compared with state-of-the-arts. The code is released at https://github.com/jliu4ai/CaPo.

Large language models (LLMs) often produce errors, including factual inaccuracies, biases, and reasoning failures, collectively referred to as "hallucinations". Recent studies have demonstrated that LLMs' internal states encode information regarding the truthfulness of their outputs, and that this information can be utilized to detect errors. In this work, we show that the internal representations of LLMs encode much more information about truthfulness than previously recognized. We first discover that the truthfulness information is concentrated in specific tokens, and leveraging this property significantly enhances error detection performance. Yet, we show that such error detectors fail to generalize across datasets, implying that---contrary to prior claims---truthfulness encoding is not universal but rather multifaceted. Next, we show that internal representations can also be used for predicting the types of errors the model is likely to make, facilitating the development of tailored mitigation strategies. Lastly, we reveal a discrepancy between LLMs' internal encoding and external behavior: they may encode the correct answer, yet consistently generate an incorrect one. Taken together, these insights deepen our understanding of LLM errors from the model's internal perspective, which can guide future research on enhancing error analysis and mitigation.