ICML 2026 论文列表

Long-horizon dialogue agents suffer from latent state drift: what an agent says, what it internally represents, and what it stores in memory can diverge silently across turns. This creates asymmetric rupture risk—many locally coherent exchanges undone by a single high-cost contradiction. We propose BRIDGE (Behavioral Reasoning through Integrated Dynamic Gated Evolution), which performs triangular fixed-point refinement to explicitly couple Observable ($\mathcal{O}$), Latent ($\mathcal{L}$), and Memory ($\mathcal{M}$) before decoding each response. We prove that under mild conditions, the refinement operator converges to a unique fixed point, providing a theoretical guarantee that the agent's internal state remains self-consistent before each response. Empirically, BRIDGE achieves the highest scores on both PersonaGym (4.59 avg., surpassing Claude-3.7-Sonnet) and CoSER (59.5% avg., +3.1 over Claude-3.7-Sonnet), with gains concentrated on persona-specific metrics (+8.0 Character Fidelity over Qwen2.5-32B-Instruct)—while updating only 0.85% trainable parameters of the frozen backbone. We also provide a Lyapunov-style uniform drift bound for tiered memory updates, grounding bounded persona evolution in long-horizon interaction.

Recent Large Multimodal Models (LMMs), primarily designed for offline settings, are ill-suited for the dynamic requirements of streaming video. While recent online adaptations improve real-time processing, they still face critical scalability challenges, with resource demands typically growing at least linearly with video duration. To overcome this bottleneck, we propose FlowNar, a novel framework for scalable streaming video narration. The core of FlowNar is a dynamic context management strategy for historical visual context removal, combined with our CLAM (Cross Linear Attentive Memory) module for streaming visual history retention, ensuring bounded visual memory usage and computational complexity, crucial for efficient streaming. We also introduce a realistic self-conditioned evaluation protocol and complementary evaluation metrics to assess streaming narration models under deployment-like conditions. Experiments on the Ego4D, EgoExo4D, and EpicKitchens100 datasets demonstrate that FlowNar substantially improves narration quality over strong baselines while being highly efficient, supporting processing of 10$\times$ longer videos and achieving 3$\times$ higher throughput (FPS). The code is available at https://github.com/zeyun-zhong/FlowNar.

Multimodal representation learning aims to construct a shared embedding space in which heterogeneous modalities are semantically aligned. Despite strong empirical results, InfoNCE-based objectives introduce inherent conflicts that yield distribution gaps across modalities. In this work, we identify two conflicts in the multimodal regime, both exacerbated as the number of modalities increases: (i) an alignment–uniformity conflict, whereby the repulsion of uniformity undermines pairwise alignment, and (ii) an intra-alignment conflict, where aligning multiple modalities induces competing alignment directions. To address these issues, we propose a principled decoupling of alignment and uniformity for multimodal representations, providing a conflict-free recipe for multimodal learning that simultaneously supports discriminative and generative use cases without task-specific modules. We then provide a theoretical guarantee that our method acts as an efficient proxy for a global Hölder divergence over multiple modality distributions, and thus reduces the distribution gap among modalities. Extensive experiments on retrieval and UnCLIP-style generation demonstrate consistent gains.

Recently, vision-language models have demonstrated increasing influence in morally sensitive domains such as autonomous driving and medical analysis, owing to their powerful multimodal reasoning capabilities. As these models are deployed in high-stakes real-world applications, it is of paramount importance to ensure that their outputs align with human moral values and remain within moral boundaries. However, existing work on moral alignment either focuses solely on textual modalities or relies heavily on AI-generated images, leading to distributional biases and reduced realism. To overcome these limitations, we introduce MORALISE, a comprehensive benchmark for evaluating the \underline{mor}al \underline{al}ignment of v\underline{is}ion-languag\underline{e} models (VLMs) using diverse, expert-verified real-world data. We begin by proposing a comprehensive taxonomy of 13 moral topics grounded in Turiel's Domain Theory, spanning the personal, interpersonal, and societal moral domains encountered in everyday life. Built on this framework, we manually curate 2,481 high-quality image-text pairs, each annotated with two fine-grained labels: (1) \textit{topic annotation}, identifying the violated moral topic(s), and (2) \textit{modality annotation}, indicating whether the violation arises from the image or the text. For evaluation, we encompass two tasks, \textit{moral judgment} and \textit{moral norm attribution}, to assess models' awareness of moral violations and their reasoning ability on morally salient content. Extensive experiments on 19 popular open- and closed-source VLMs show that MORALISE poses a significant challenge, revealing persistent moral limitations in current state-of-the-art models.

Aligning Multimodal Large Language Models (MLLMs) with human preferences remains a fundamental challenge. While Generative Reward Models (GRMs) offer a promising reasoning-based alternative to scalar models, they are often hindered by severe position bias and prohibitively high computational overhead. To address these limitations, we propose GenAlign, a unified framework that synergizes robust generative reward modeling with efficient MLLM alignment. First, we introduce a rubric-based GRM that explicitly models the preference judgment process. By employing reinforcement learning with verifiable rewards and an online position debiasing mechanism, our model produces interpretable reasoning critiques and robust preference predictions. Second, we propose a policy optimization strategy utilizing advantage-smoothed dynamic reference anchoring. This approach reduces computational complexity while mitigating gradient instability caused by variance collapse. Extensive experiments demonstrate that GenAlign achieves state-of-the-art preference prediction accuracy on multimodal reward modeling benchmarks. Moreover, it consistently improves the performance of three MLLMs across seven diverse evaluation benchmarks, particularly making significant progress in safety and hallucination.

Vision-language models (VLMs) now support both direct Instruct and explicit-reasoning Thinking modes, yet practitioners still lack principled ways to decide when reasoning actually improves performance, or how much computation to spend at test time, so we investigate whether VLMs encode meta-cognitive signals for adaptive inference. We derive oracle labels for two properties: (1) reasoning helpfulness, namely whether explicit reasoning improves accuracy, and (2) desired generation length, the minimal token budget needed for a correct answer. Probing final-layer representations in InternVL and Qwen3-VL models, we find Thinking models encode these signals more linearly than Instruct models, suggesting that reasoning-oriented training enhances meta-cognitive structure. Head-wise attribution reveals two circuits: length-control heads in lower layers and reasoning/difficulty heads in higher layers. Causal interventions confirm these roles across benchmarks: scaling length heads controls output length with minimal accuracy loss, while scaling reasoning heads enables a perception-reasoning trade-off, improving accuracy by up to 5.3\%. Our results demonstrate that reasoning-tuned VLMs develop localized, manipulable circuits for meta-cognitive control, enabling test-time steering of computation and reasoning without retraining.

We introduce Alterbute, a diffusion-based method for editing an object's intrinsic attributes in an image. We allow changing color, texture, material, and even the shape of an object, while preserving its perceived identity and scene context. Existing approaches either rely on unsupervised priors that often fail to preserve identity or use overly restrictive supervision that prevents meaningful intrinsic variations. Our method relies on: (i) a relaxed training objective that allows the model to change both intrinsic and extrinsic attributes conditioned on an identity reference image, a textual prompt describing the target intrinsic attributes, and a background image and object mask defining the extrinsic context. At inference, we restrict extrinsic changes by reusing the original background and object mask, thereby ensuring that only the desired intrinsic attributes are altered; (ii) Visual Named Entities (VNEs) - fine-grained visual identity categories (e.g., "Porsche 911 Carrera") that group objects sharing identity-defining features while allowing variation in intrinsic attributes. We use a vision-language model to automatically extract VNE labels and intrinsic attribute descriptions from a large public image dataset, enabling scalable, identity-preserving supervision. Alterbute outperforms existing methods on identity-preserving object intrinsic attribute editing.

Despite recent Multimodal Large Language Models (MLLMs)' linguistic prowess in medical diagnosis, we find even state-of-the-art MLLMs suffer from a critical perceptual deficit: **geometric blindness**. This failure to ground outputs in objective geometric constraints leads to plausible yet factually incorrect hallucinations, rooted in training paradigms that prioritize linguistic fluency over geometric fidelity. This paper introduces Med-Scout, a novel framework that "cures" this blindness via Reinforcement Learning (RL) that leverages the intrinsic geometric logic latent within unlabeled medical images. Instead of relying on costly expert annotations, Med-Scout derives verifiable supervision signals through three strategic proxy tasks inspired by the systematic reading and reasoning patterns of clinicians: Hierarchical Scale Localization, Topological Jigsaw Reconstruction, and Anomaly Consistency Detection. To rigorously quantify this deficit, we present Med-Scout-Bench, a new benchmark specifically designed to evaluate geometric perception. Extensive evaluations show that Med-Scout significantly mitigates geometric blindness, outperforming leading proprietary and open-source MLLMs by over **40%** on our benchmark. Furthermore, this enhanced geometric perception generalizes to broader medical understanding, achieving superior results on radiological and comprehensive medical VQA tasks. The project page is: https://github.com/HKUSTGZ-ML4Health-Lab/Med-Scout.

Hallucination is a central limitation of large language models (LLMs), and substantial effort has been devoted to understanding and mitigating it. Towards this, Kalai and Vempala (STOC 2024) introduced a probabilistic framework formalizing calibration and hallucination, and showed that, with high probability, calibrated LLMs hallucinate roughly at the rate of the "missing mass", a measure of how incomplete the training data is relative to its source. This raises two fundamental questions: (i) what property of a calibrated LLM makes hallucinations unavoidable? and (ii) can hallucinations be avoided by giving up calibration? We answer these questions by introducing a simpler property we call *innovation* that measures the tendency of a model to produce outputs outside the training data. We show that innovation is implied by the condition for hallucination identified by Kalai and Vempala, and, further, that it is an almost characterization of hallucination: hallucination implies innovation, and conversely, innovation implies hallucination with high probability. We also provide lower bounds on the hallucination rate based on the "innovation rate", and by relating innovation rate back to missing mass, we obtain new hallucination rate lower bounds based on missing mass that extend the results of Kalai and Vempala.

Visual prostheses hold great promise for restoring vision in blind individuals. While researchers have successfully utilized M/EEG signals to evoke visual perceptions during the brain decoding stage of visual prostheses, the complementary process of converting images into M/EEG signals in the brain encoding stage remains largely unexplored, hindering the formation of a complete functional pipeline. In this work, we present a novel image-to-brain signal framework that generates M/EEG from images by leveraging the diffusion transformer architecture enhanced with cross-attention mechanisms. Specifically, we employ a diffusion transformer (DiT) architecture based on denoising diffusion implicit models (DDIM) to achieve brain signal generation. To realize the goal of image-to-brain signal conversion, we use cross-attention mechanisms to align brain signal embeddings with CLIP image embeddings. Moreover, we leverage large language models (LLMs) to generate image captions, and concatenate the resulting CLIP text embeddings with CLIP image embeddings to form unified embeddings for cross-attention alignment, enabling our model to capture core semantic information. Furthermore, we introduce a learnable spatio-temporal position encoding that combines brain region embeddings with temporal embeddings to capture both spatial and temporal characteristics of brain signals. We evaluate the framework on two multimodal benchmark datasets (THINGS-EEG2 and THINGS-MEG) and demonstrate that it generates biologically plausible brain signals.

Standard factuality evaluations of LLMs treat all errors alike, obscuring whether failures arise from missing knowledge (empty shelves) or from limited access to encoded facts (lost keys). We propose a behavioral framework that profiles factual knowledge at the level of facts rather than questions, characterizing each fact by whether it is encoded, and then by how accessible it is: cannot be recalled, can be directly recalled, or can only be recalled with inference-time computation (thinking). To support such profiling, we introduce WikiProfile, a new benchmark constructed via an automated pipeline with a prompted LLM grounded in web search. Across 4 million responses from 13 LLMs, we find that encoding is nearly saturated in frontier models on our benchmark, with GPT-5 and Gemini-3 encoding 95--98\% of facts. However, recall remains a major bottleneck: many errors previously attributed to missing knowledge instead stem from failures to access it. These failures are systematic and disproportionately affect long-tail facts and reverse questions. Finally, we show that thinking improves recall and can recover a substantial fraction of failures, indicating that future gains may rely less on scaling and more on methods that improve how models utilize what they already encode.

Existing EEG models are limited by electrode heterogeneity and rigid "channel-first" architectures that treat sensors as independent features. We propose Brain Signal Rendering (BSR), which reinterprets EEG as a physical projection of neural activity and transforms raw signals into structured spatiotemporal tensors (termed Spectrum Videos), enabling the transfer of rich priors from video foundation models. By utilizing VideoMAE for self-supervised pre-training, BSR learns robust, layout-agnostic spatiotemporal representations that preserve neural topology. We further employ subject-level few-shot learning and introduce cross-montage fine-tuning to rigorously evaluate generalization across subjects and electrode configurations. Experiments show that VideoMAE model integrated with the BSR framework significantly outperforms state-of-the-art spectrum based methods, providing a scalable and data-efficient foundation for generalizable EEG modeling. Our code is available at [https://github.com/yanweifu-sii/BSR-VideoMAE](https://github.com/yanweifu-sii/BSR-VideoMAE).

Recent advances in Vision-Language-Action (VLA) models have significantly improved robotic perception and manipulation capabilities, but still struggling to adapt in dynamic, open-ended real-world environments due to a lack of reliable task progress feedback and improvement mechanisms. To address these challenges, we propose a generalist Vision Language Action-Critic model, VLAC, which can integrate both human and robot data, and unify action policy and task progress critic within a single autoregressive architecture. Specifically, we propose a scalable and generalizable pair-wise progress understanding approach that can predict the delta of task progress between two steps in a trajectory and generate correct actions to complete the task. Then, we trained the model on large-scale, multi-source human, robot, and general vision-language data for a generalist. Furthermore, we deploy reinforcement learning where VLAC can autonomously evaluate task progress to provide intrinsic rewards. Extensive evaluations demonstrate that our model generalizes effectively across diverse tasks and environments, leveraging its pair-wise progress understanding to provide reliable dense rewards, robust action generation, and significant improvements in real-world reinforcement learning. Our codes are available at \url{https://github.com/InternRobotics/VLAC}.

Developing effective representations for register transfer level (RTL) designs is crucial for accelerating the hardware design workflow. Existing approaches, however, typically rely on a single data modality, either the RTL code or its associated graph-based representation, limiting the expressiveness and generalization ability of the learned representations. For RTL, the control data flow graph (CDFG) offers a comprehensive structural representation that preserves complete information, while the code modality explicitly encodes semantic and functional information. We argue that integrating these complementary modalities is essential for a thorough understanding of RTL designs. To this end, we propose UniRTL, a multimodal pretraining framework that learns unified RTL representations by jointly leveraging code and CDFG. UniRTL achieves fine-grained alignment between code and graph through mutual masked modeling and employs a hierarchical training strategy that incorporates a pretrained graph-aware tokenizer and staged alignment of text (*i.e.*, functional summary) and code prior to graph integration. We evaluate UniRTL on two downstream tasks, performance prediction and code retrieval, under multiple settings. Experimental results show that UniRTL consistently outperforms prior methods, establishing it as a more robust and powerful foundation for advancing hardware design automation.

Incorporating group symmetries via equivariance into neural networks has emerged as a robust approach for improving efficiency and overcoming the data requirements of modern deep learning. While most existing approaches, such as group convolutions and averaging-based methods, focus on compact, finite, or low-dimensional groups with linear actions, this work explores how equivariance can be extended to infinite-dimensional groups. We propose a strategy designed to induce diffeomorphism equivariance in pre-trained neural networks via energy-based canonicalization. Formulating equivariance as an optimization problem allows us to access the rich toolbox of already established differentiable image registration methods. Empirical results on segmentation and classification tasks confirm that our approach achieves approximate equivariance and generalizes to unseen transformations without relying on extensive data augmentation or retraining.

Temporal Grounding (TG) aims to localize video segments corresponding to a textual query. Prior research predominantly focuses on single-segment retrieval. Real-world scenarios, however, often require localizing multiple disjoint segments for a single query—a setting we term **One-to-Many Temporal Grounding (OMTG)**. Previous state-of-the-art MLLMs, optimized for one-to-one settings, struggle in this context, often yielding near-zero scores due to a lack of event cardinality perception. To bridge this gap, we present a systematic solution with three key contributions. First, we establish the first comprehensive OMTG benchmark, introducing Count Accuracy (C-Acc) and Effective Temporal F1 (EtF1) as evaluation metrics. Second, we curate a high-quality OMTG dataset comprising 56k samples through a sophisticated construction pipeline. Third, we develop novel temporal and caption reward functions specifically designed for OMTG. In particular, the caption reward leverages Chain-of-Thought reasoning over dense video captions to explicitly guide policy optimization toward both preciseness and completeness. Extensive experiments show our model achieves a new state-of-the-art EtF1 of 43.65\% on OMTG Bench, outperforming Gemini 2.5 Pro and Seed-1.8 by 15.85\% and 15.61\%, respectively.

Reasoning segmentation is an emerging vision-language task that requires reasoning over intricate text queries to precisely segment objects. However, existing methods typically suffer from overthinking, generating verbose reasoning chains that interfere with object localization in multimodal large language models (MLLMs). To address this issue, we propose DR$^2$Seg, a self-rewarding framework that improves both reasoning efficiency and segmentation accuracy without requiring extra thinking supervision. DR$^2$Seg employs a two-stage rollout strategy that decomposes reasoning segmentation into multimodal reasoning and referring segmentation. In the first stage, the model generates a self-contained description that explicitly specifies the target object. In the second stage, this description replaces the original complex query to verify its self-containment. Based on this design, two self-rewards are introduced to mitigate overthinking and the associated attention dispersion. Extensive experiments conducted on 3B and 7B variants of Qwen2.5-VL, as well as on both SAM2 and SAM3, demonstrate that DR$^2$Seg consistently improves reasoning efficiency and overall segmentation accuracy. The source code can be found at https://github.com/harrylin-hyl/DR2Seg.

Recent advances in large language models (LLMs) have led to remarkable progress across various domains, yet their capabilities in the humanities, particularly history, remain underexplored. Historical reasoning poses unique challenges for LLMs, involving multimodal source interpretation, temporal inference, and cross-linguistic analysis. Existing general-purpose agents perform well on many current benchmarks but lack the domain expertise needed to address complex historical questions. To address this gap, we introduce HistBench, a new benchmark of 414 high-quality and carefully-reviewed questions stratified by difficulty and designed to evaluate LLM's capacity for historical reasoning. The tasks span a wide range of historical problems—from factual retrieval based on primary sources to interpretive analysis of manuscripts and images, to interdisciplinary challenges involving archaeology, linguistics, or cultural history. Furthermore, the benchmark dataset spans 29 ancient and modern languages and covers a wide range of historical periods and world regions. Finding the poor performance of LLMs and other agents on HistBench, we further present HistAgent, a history-specific agent equipped with carefully designed tools for OCR, translation, archival search, and image understanding in history. On HistBench, HistAgent based on GPT-4o achieves an accuracy of 28.50\% pass@1 and 36.47\% pass@2, significantly outperforming LLMs with online search and generalist agents, including GPT-4o (18.60\%), DeepSeek-R1 (14.98\%), Grok 3 (17.63\%) and Open Deep Research by smolagents (19.57\% pass@1 and 25.12\% pass@2). These results highlight the limitations of existing LLMs and generalist agents and demonstrate the advantages of HistAgent for historical reasoning. Notably, HistAgent also achieves 60.00\% pass@1 accuracy on the GAIA benchmark, showing that domain-specific customization doesn't hinder HistAgent's competitive performance on real-world general tasks. Code is available at https://github.com/CharlesQ9/HistAgent.

Group Emotion Recognition (GER) is crucial for understanding social dynamics, ranging from interpreting intimate conversations to evaluating crowd behavior in large-scale surveillance scenarios. While current AI models can analyze these scenes, they often act as black boxes that take shortcuts. Instead of focusing on how people are actually behaving, these models often get distracted by the background environment, leading to inaccurate results. To bridge this gap, we introduce VIBE (Variational Inference for Behavioral Emotion), a kinematics-aware framework that integrates audio, video, and text modalities through causal structuring. Unlike standard models that simply mix data together, VIBE utilizes mathematical constraints to filter out background noise and isolate the genuine emotions of the people involved. This purified representation enables our model to focus exclusively on the sociological mechanics of the crowd, dynamically modulating neural attention based on raw physical synchrony. Simultaneously, we align visual dynamics with human interpretability by projecting latent representations into a semantically structured space informed by textual descriptions. Comprehensive experiments demonstrate that VIBE consistently outperforms state-of-the-art methods. Code is available at GitHub.