Publications

@conference{nokeyi,

title = {FaithCoT-Bench: Benchmarking Instance-Level Faithfulness of Chain-of-Thought Reasoning},

author = {Xu Shen and Song Wang and Zhen Tan and Laura Yao and Xinyu Zhao and Kaidi Xu and Xin Wang and Tianlong Chen

},

url = {https://arxiv.org/abs/2510.04040},

year  = {2026},

date = {2026-01-29},

urldate = {2026-01-29},

booktitle = {Fourteenth International Conference on Learning Representations (ICLR)},

abstract = {Large language models (LLMs) increasingly rely on Chain-of-Thought (CoT) prompting to improve problem-solving and provide seemingly transparent explanations. However, growing evidence shows that CoT often fail to faithfully represent the underlying reasoning process, raising concerns about their reliability in high-risk applications. Although prior studies have focused on mechanism-level analyses showing that CoTs can be unfaithful, they leave open the practical challenge of deciding whether a specific trajectory is faithful to the internal reasoning of the model. To address this gap, we introduce FaithCoT-Bench, a unified benchmark for instance-level CoT unfaithfulness detection. Our framework establishes a rigorous task formulation that formulates unfaithfulness detection as a discriminative decision problem, and provides FINE-CoT (Faithfulness instance evaluation for Chain-of-Thought), an expert-annotated collection of over 1,000 trajectories generated by four representative LLMs across four domains, including more than 300 unfaithful instances with fine-grained causes and step-level evidence. We further conduct a systematic evaluation of eleven representative detection methods spanning counterfactual, logit-based, and LLM-as-judge paradigms, deriving empirical insights that clarify the strengths and weaknesses of existing approaches and reveal the increased challenges of detection in knowledge-intensive domains and with more advanced models. To the best of our knowledge, FaithCoT-Bench establishes the first comprehensive benchmark for instance-level CoT faithfulness,},

keywords = {ICLR},

pubstate = {published},

tppubtype = {conference}

}

@conference{nokeyj,

title = {RewardMap: Tackling Sparse Rewards in Fine-grained Visual Reasoning via Multi-Stage Reinforcement Learning},

author = {Sicheng Feng and Kaiwen TUO and Song Wang and Lingdong Kong and Jianke Zhu and Huan Wang

},

url = {https://arxiv.org/abs/2510.02240, https://fscdc.github.io/RewardMap/},

year  = {2026},

date = {2026-01-26},

urldate = {2026-01-29},

booktitle = {Fourteenth International Conference on Learning Representations (ICLR)},

abstract = {Distilling robust reasoning capabilities from large language models (LLMs) into smaller, computationally efficient student models remains an unresolved challenge. Despite recent advances, distilled models frequently suffer from superficial pattern memorization and subpar generalization. To overcome these limitations, we introduce a novel distillation framework that moves beyond simple mimicry to instill a deeper conceptual understanding. Our framework features two key innovations. underline{textit{First}}, to address pattern memorization, Explanatory Inversion (EI) generates targeted ``explanatory probes'' that compel the student to articulate the underlying logic behind an answer, rather than just memorizing it. underline{textit{Second}}, to improve generalization, Explanatory GRPO (texttt{EXGRPO}) uses a reinforcement learning algorithm with a novel Dialogue Structure Utility Bonus, which explicitly rewards the student for maintaining a coherent reasoning process across these probes. Extensive evaluations on 12 datasets demonstrate significant improvements. Using Gemma-7b as the student model, our method yields an average textbf{20.39%} increase over zero-shot performance and a textbf{6.02%} improvement over the state-of-the-art distillation baselines. Moreover, models distilled with our method show remarkable training efficiency (e.g., surpassing vanilla fine-tuning with textbf{10-25%} training data) and strong generalization to out-of-distribution tasks.},

keywords = {ICLR},

pubstate = {published},

tppubtype = {conference}

}

@conference{nokeyk,

title = {Probing to Refine: Reinforcement Distillation of LLM Reasoners via Explanatory Inversion},

author = {Zhen Tan and Chengshuai Zhao and Song Wang and Jundong Li and Tianlong Chen and huan liu

},

url = {https://openreview.net/forum?id=rkIw2GqYEt},

year  = {2026},

date = {2026-01-26},

urldate = {2026-01-26},

booktitle = {The Fourteenth International Conference on Learning Representations},

abstract = {Distilling robust reasoning capabilities from large language models (LLMs) into smaller, computationally efficient student models remains an unresolved challenge. Despite recent advances, distilled models frequently suffer from superficial pattern memorization and subpar generalization. To overcome these limitations, we introduce a novel distillation framework that moves beyond simple mimicry to instill a deeper conceptual understanding. Our framework features two key innovations. underline{textit{First}}, to address pattern memorization, Explanatory Inversion (EI) generates targeted ``explanatory probes'' that compel the student to articulate the underlying logic behind an answer, rather than just memorizing it. underline{textit{Second}}, to improve generalization, Explanatory GRPO (texttt{EXGRPO}) uses a reinforcement learning algorithm with a novel Dialogue Structure Utility Bonus, which explicitly rewards the student for maintaining a coherent reasoning process across these probes. Extensive evaluations on 12 datasets demonstrate significant improvements. Using Gemma-7b as the student model, our method yields an average textbf{20.39%} increase over zero-shot performance and a textbf{6.02%} improvement over the state-of-the-art distillation baselines. Moreover, models distilled with our method show remarkable training efficiency (e.g., surpassing vanilla fine-tuning with textbf{10-25%} training data) and strong generalization to out-of-distribution tasks.},

keywords = {ICLR},

pubstate = {published},

tppubtype = {conference}

}

@conference{nokeyl,

title = {VisionTrim: Unified Vision Token Compression for Training-Free MLLM Acceleration},

author = {Hanxun Yu and Wentong Li and Xuan Qu and Song Wang and Junbo Chen and Jianke Zhu

},

url = {https://arxiv.org/abs/2601.22674},

year  = {2026},

date = {2026-01-26},

booktitle = {Fourteenth International Conference on Learning Representations (ICLR), 2026.},

abstract = {Multimodal large language models (MLLMs) suffer from high computational costs due to excessive visual tokens, particularly in high-resolution and video-based scenarios. Existing token reduction methods typically focus on isolated pipeline components and often neglect textual alignment, leading to performance degradation. In this paper, we propose VisionTrim, a unified framework for training-free MLLM acceleration, integrating two effective plug-and-play modules: 1) the Dominant Vision Token Selection (DVTS) module, which preserves essential visual tokens via global-local view, and 2) the Text-Guided Vision Complement (TGVC) module, which facilitates context-aware token merging guided by textual cues. Extensive experiments across diverse image and video multimodal benchmarks demonstrate the performance superiority of our VisionTrim, advancing practical MLLM deployment in real-world applications. Our full implementation will be publicly available.},

keywords = {ICLR},

pubstate = {published},

tppubtype = {conference}

}

@conference{nokeym,

title = {A.I.R.: Enabling Adaptive, Iterative, and Reasoning-based Frame Selection For Video Question Answering},

author = {Yuanhao Zou and Shengji Jin and Andong Deng and Youpeng Zhao and Jun Wang and Chen Chen

},

url = {https://ucf-air.github.io/},

year  = {2026},

date = {2026-01-26},

urldate = {2026-01-26},

booktitle = {The Fourteenth International Conference on Learning Representations},

abstract = {Effectively applying Vision-Language Models (VLMs) to Video Question Answering (VideoQA) hinges on selecting a concise yet comprehensive set of frames, as processing entire videos is computationally infeasible. However, current frame selection methods face a critical trade-off: approaches relying on lightweight similarity models, such as CLIP, often fail to capture the nuances of complex queries, resulting in inaccurate similarity scores that cannot reflect the authentic query-frame relevance, which further undermines frame selection. Meanwhile, methods that leverage a VLM for deeper analysis achieve higher accuracy but incur prohibitive computational costs. To address these limitations, we propose A.I.R., a training-free approach for Adaptive, Iterative, and Reasoning-based frame selection. We leverage a powerful VLM to perform deep, semantic analysis on complex queries, and this analysis is deployed within a cost-effective iterative loop that processes only a small batch of the most high-potential frames at a time. Extensive experiments on various VideoQA benchmarks demonstrate that our approach outperforms existing frame selection methods, significantly boosts the performance of the foundation VLM, and achieves substantial gains in computational efficiency over other VLM-based techniques.},

keywords = {ICLR},

pubstate = {published},

tppubtype = {conference}

}

@conference{nokeyn,

title = {FlowNIB: An Information Bottleneck Analysis of Bidirectional vs. Unidirectional Language Models},

author = {Md Kowsher and Nusrat Prottasha and Shiyun Xu and Shetu Mohanto and OZLEM GARIBAY and Niloofar Yousefi and Chen Chen

},

url = {https://github.com/Kowsher/BidiVsUniLM},

year  = {2026},

date = {2026-01-26},

booktitle = {The Fourteenth International Conference on Learning Representations},

abstract = {Bidirectional language models (LMs) consistently show stronger context understanding than unidirectional models, yet the theoretical reason remains unclear. We present a simple information bottleneck (IB) perspective: bidirectional representations preserve more mutual information (MI) about both the input and the target, yielding richer features for downstream tasks. We adopt a layer–wise view and hypothesize that, at comparable capacity, bidirectional layers retain more useful signal than unidirectional ones. To test this claim empirically, we present Flow Neural Information Bottleneck (FlowNIB), a lightweight, post-hoc framework capable of estimating comparable mutual information values for individual layers in LMs, quantifying how much mutual information each layer carries for a dataset. FlowNIB takes three inputs—(i) the original LM’s inputs/dataset, (ii) ground–truth labels, and (iii) layer activations—simultaneously estimates the mutual information for both the input–layer and layer–label pairs. Empirically, bidirectional LM layers exhibit higher mutual information than similar—and even larger—unidirectional LMs. As a result, bidirectional LMs outperform unidirectional LMs across extensive experiments on NLU benchmarks (e.g., GLUE), commonsense reasoning, and regression tasks, demonstrating superior context understanding.},

keywords = {ICLR},

pubstate = {published},

tppubtype = {conference}

}

@conference{nokeyp,

title = {ViPO: Visual Preference Optimization at Scale},

author = {Ming Li and Jie Wu and Jiaxing Cui and Xiaojie Li and Rui Wang and Chen Chen

},

url = {https://openreview.net/pdf?id=x5zP3k64Nl},

year  = {2026},

date = {2026-01-26},

booktitle = {The Fourteenth International Conference on Learning Representations},

abstract = {While preference optimization is crucial for improving visual generative models, how

to effectively scale this paradigm for visual generation remains largely unexplored.

Current open-source preference datasets typically contain substantial conflicting

preference patterns, where winners excel in some dimensions but underperform in

others. Naively optimizing on such noisy datasets fails to learn meaningful preferences,

fundamentally hindering effective scaling. To enhance the robustness of preference

algorithms against noise, we propose Poly-DPO, which extends the DPO objective

with an additional polynomial term that dynamically adjusts model confidence during

training based on dataset characteristics, enabling effective learning across diverse

data distributions from noisy to trivially simple patterns. Beyond biased patterns,

existing datasets suffer from low resolution, limited prompt diversity, and imbalanced

distributions. To facilitate large-scale visual preference optimization by tackling key

data bottlenecks, we construct ViPO, a massive-scale preference dataset with 1M

image pairs (1024px) across five categories and 300K video pairs (720p+) across three

categories. Leveraging state-of-the-art generative models and diverse prompts ensures

consistent, reliable preference signals with balanced distributions. Remarkably, when

applying Poly-DPO to our high-quality dataset, the optimal configuration converges

to standard DPO. This convergence validates both our dataset quality and Poly-DPO’s

adaptive nature: sophisticated optimization becomes unnecessary with sufficient data

quality, yet remains valuable for imperfect datasets. We comprehensively validate our

approach across various visual generation models. On noisy datasets like Pick-a-Pic V2,

Poly-DPO achieves 6.87 and 2.32 gains over Diffusion-DPO on GenEval for SD1.5 and

SDXL, respectively. For our high-quality ViPO dataset, models achieve performance

far exceeding those trained on existing open-source preference datasets. These results

confirm that addressing both algorithmic adaptability and data quality is essential for

scaling visual preference optimization. All models and datasets will be released.},

keywords = {ICLR},

pubstate = {published},

tppubtype = {conference}

}

@conference{nokeyq,

title = {Implicit Models: Expressive Power Scales with Test-Time Compute},

author = {Jialin Liu and Lisang Ding and Wotao Yin and Stanley J Osher

},

url = {https://arxiv.org/abs/2510.03638},

year  = {2026},

date = {2026-01-26},

booktitle = {The Fourteenth International Conference on Learning Representations},

abstract = {Implicit models, an emerging model class, compute outputs by iterating a single parameter block to a fixed point. This architecture realizes an infinite-depth, weight-tied network that trains with constant memory, significantly reducing memory needs for the same level of performance compared to explicit models. While it is empirically known that these compact models can often match or even exceed the accuracy of larger explicit networks by allocating more test-time compute, the underlying reasons are not yet well understood. We study this gap through a non-parametric analysis of expressive power. We provide a strict mathematical characterization, showing that a simple and regular implicit operator can, through iteration, progressively express more complex mappings. We prove that for a broad class of implicit models, this process allows the model's expressive power to grow with test-time compute, ultimately matching a much richer function class. The theory is validated across four domains: imaging, scientific computing, operations research, and LLM reasoning, demonstrating that as test-time iterations increase, the complexity of the learned mapping rises, while the solution quality simultaneously improves and stabilizes},

keywords = {ICLR},

pubstate = {published},

tppubtype = {conference}

}

@conference{nokeyr,

title = {Understanding and improving Shampoo and SOAP via Kullback-Leibler Minimization},

author = {Wu Lin and Scott C Lowe and Felix Dangel and Runa Eschenhagen and Zikun Xu and Roger Grosse

},

url = {https://arxiv.org/abs/2509.03378},

year  = {2026},

date = {2026-01-26},

urldate = {2026-01-26},

booktitle = {The Fourteenth International Conference on Learning Representations},

abstract = {Shampoo and its efficient, Adam-stabilized variant SOAP, employ structured second-moment estimation and have received growing attention for their effectiveness. In practice, Shampoo requires step-size grafting with Adam to achieve competitive performance. SOAP mitigates this by applying Adam in Shampoo's eigenbasis and further reducing per-iteration runtime. However, reliance on Adam introduces additional memory overhead in both methods. Prior theoretical interpretations have primarily examined their estimation schemes using the Frobenius norm. Motivated by the natural correspondence between the second moment and a covariance matrix, we reinterpret the estimation procedures in Shampoo and SOAP as instances of covariance estimation through the lens of Kullback–Leibler (KL) divergence minimization. This perspective reveals a previously overlooked theoretical limitation and motivates principled improvements to their design. Building on the KL perspective, we propose practical estimation schemes---KL-Shampoo and KL-SOAP---that match or exceed the performance of Shampoo and SOAP for pre-training a range of neural network models while maintaining SOAP-level per-iteration runtime. Notably, KL-Shampoo does not rely on Adam to achieve superior performance, thereby avoiding the associated memory overhead. Surprisingly, KL-Shampoo consistently outperforms the other methods in our experiments.},

keywords = {ICLR},

pubstate = {published},

tppubtype = {conference}

}

@conference{nokeys,

title = {AC-Foley: Reference-Audio-Guided Video-to-Audio Synthesis with Acoustic Transfer},

author = {Pengjun Fang and Yingqing He and Yazhou Xing and Qifeng Chen and Ser-Nam Lim and Harry Yang

},

url = {https://arxiv.org/abs/2603.15597},

year  = {2026},

date = {2026-01-26},

urldate = {2026-01-26},

booktitle = {The Fourteenth International Conference on Learning Representations},

abstract = {Existing video-to-audio (V2A) generation methods predominantly rely on text prompts alongside visual information to synthesize audio. However, two critical bottlenecks persist: semantic granularity gaps in training data (e.g., conflating acoustically distinct sounds like different dog barks under coarse labels), and textual ambiguity in describing microacoustic features (e.g., "metallic clang" failing to distinguish impact transients and resonance decay). These bottlenecks make it difficult to perform fine-grained sound synthesis using text-controlled modes. To address these limitations, we propose AC-Foley, an audio-conditioned V2A model that directly leverages reference audio to achieve precise and fine-grained control over generated sounds. This approach enables: fine-grained sound synthesis (e.g., footsteps with distinct timbres on wood, marble, or gravel), timbre transfer (e.g., transforming a violin’s melody into the bright, piercing tone of a suona), zero-shot generation of sounds (e.g., creating unique weapon sound effects without training on firearm datasets) and better audio quality. By directly conditioning on audio signals, our approach bypasses the semantic ambiguities of text descriptions while enabling precise manipulation of acoustic attributes. Empirically, AC-Foley achieves state-of-the-art performance for Foley generation when conditioned on reference audio, while remaining competitive with SOTA video-to-audio methods even without audio conditioning.},

keywords = {ICLR},

pubstate = {published},

tppubtype = {conference}

}

@conference{nokeyt,

title = {Think Then Embed: Generative Context Improves Multimodal Embedding},

author = {Xuanming Cui and Jianpeng Cheng and Hong-You Chen and Satya Narayan Shukla and Abhijeet Awasthi and Xichen Pan and Chaitanya Ahuja and Shlok Mishra and Taipeng Tian and Qi Guo and Ser-Nam Lim and Aashu Singh and Xiangjun Fan},

url = {https://arxiv.org/pdf/2510.05014},

year  = {2026},

date = {2026-01-26},

booktitle = {The Fourteenth International Conference on Learning Representations},

abstract = {There is a growing interest in Universal Multimodal Embeddings (UME), where models are required to generate task-specific representations. While recent studies show that Multimodal Large Language Models (MLLMs) perform well on such tasks, they treat MLLMs solely as encoders, overlooking their generative capacity. However, such an encoding paradigm becomes less effective as instructions become more complex and require compositional reasoning. Inspired by the proven effectiveness of chain-of-thought reasoning, we propose a general Think-Then-Embed (TTE) framework for UME, composed of a reasoner and an embedder. The reasoner MLLM first generates reasoning traces that explain complex queries, followed by an embedder that produces representations conditioned on both the original query and the intermediate reasoning. This explicit reasoning step enables more nuanced understanding of complex multimodal instructions. Our contributions are threefold. First, by leveraging a powerful MLLM reasoner, we achieve state-of-the-art performance on the MMEB-V2 benchmark, surpassing proprietary models trained on massive in-house datasets. Second, to reduce the dependency on large MLLM reasoners, we finetune a smaller MLLM reasoner using high-quality embedding-centric reasoning traces, achieving the best performance among open-source models with a 7% absolute gain over recently proposed models. Third, we investigate strategies for integrating the reasoner and embedder into a unified model for improved efficiency without sacrificing performance.},

keywords = {ICLR},

pubstate = {published},

tppubtype = {conference}

}

@conference{nokeyu,

title = {Learning to Parallel: Accelerating Diffusion Large Language Models via Adaptive Parallel Decoding},

author = {Wenrui Bao and Zhiben Chen and Dan Xu and Yuzhang Shang},

url = {https://arxiv.org/abs/2509.25188},

year  = {2026},

date = {2026-01-26},

booktitle = {The Fourteenth International Conference on Learning Representations},

abstract = {Autoregressive decoding in large language models (LLMs) requires sequential steps for O(n) tokens, fundamentally limiting inference throughput. Recent diffusion-based LLMs (dLLMs) enable parallel token generation through iterative denoising. However, current parallel decoding strategies rely on fixed, input-agnostic heuristics (e.g., confidence thresholds), which fail to adapt to input-specific characteristics, resulting in suboptimal speed-quality trade-offs across diverse NLP tasks. In this work, we explore a more flexible and dynamic approach to parallel decoding. We propose **Learning to Parallel Decode (Learn2PD)**, a framework that trains a lightweight and adaptive filter model to predict, for each token position, whether the current prediction matches the final output. This learned filter approximates an oracle parallel decoding strategy that unmasks tokens only when correctly predicted. Importantly, the filter model is learned in a post-training manner, requiring only a small amount of computation to optimize it (minute-level GPU time). Additionally, we introduce **End-of-Text Prediction (EoTP)** to detect decoding completion at the end of sequence, avoiding redundant decoding of padding tokens. Experiments on the LLaDA benchmark demonstrate that our method achieves up to **22.58×** speedup without any performance drop, and up to **57.51×** when combined with KV-Cache.},

keywords = {ICLR},

pubstate = {published},

tppubtype = {conference}

}

@conference{nokeyv,

title = {Real-Time Robot Execution with Masked Action Chunking},

author = {Haoxuan Wang and Gengyu Zhang and Yan Yan and Yuzhang Shang and Ramana Kompella and Gaowen Liu

},

url = {https://arxiv.org/abs/2601.20130},

year  = {2026},

date = {2026-01-26},

booktitle = {The Fourteenth International Conference on Learning Representations

},

abstract = {Real-time execution is essential for cyber-physical systems such as robots. These systems operate in dynamic real-world environments where even small delays can undermine responsiveness and compromise performance. Asynchronous inference has recently emerged as a system-level paradigm for real-time robot manipulation, enabling the next action chunk to be predicted while the current one is being executed. While this approach achieves real-time responsiveness, naive integration often results in execution failure. Previous methods attributed this failure to inter-chunk discontinuity and developed test-time algorithms to smooth chunk boundaries. In contrast, we identify another critical yet overlooked factor: intra-chunk inconsistency, where the robot’s executed action chunk partially misaligns with its current perception. To address this, we propose REMAC, which learns corrective adjustments on the pretrained policy through masked action chunking, enabling the policy to remain resilient under mismatches between intended actions and actual execution during asynchronous inference. In addition, we introduce a prefix-preserved sampling procedure to reinforce inter-chunk continuity. Overall, our method delivers more reliable policies without incurring additional latency. Extensive experiments in both simulation and real-world settings demonstrate that our method enables faster task execution, maintains robustness across varying delays, and consistently achieves higher completion rates.},

keywords = {ICLR},

pubstate = {published},

tppubtype = {conference}

}

@conference{nokeyw,

title = {LEGACY: A Lightweight Dynamic Gradient Compression Strategy for Distributed Deep Learning},

author = {Mostapha Essoullami and El Houcine Bergou and Aritra Dutta

},

url = {https://openreview.net/forum?id=DsBYEO3B8O},

year  = {2026},

date = {2026-01-26},

urldate = {2026-01-26},

booktitle = {The Fourteenth International Conference on Learning Representations},

abstract = {Distributed learning has achieved remarkable success in training deep neural networks (DNNs) on large datasets, but the communication bottleneck limits its scalability. Various compression techniques have been proposed to alleviate this limitation; however, they either use fixed parameters throughout training or rely on complex and computationally intensive methods to adapt compression parameters. Instead of the hard-to-tune hyperparameters required by adaptive compressors, this paper investigates the impact of two fundamental factors in DNN training—the layer size of the networks and their training phases—to design a simple yet efficient dynamic scheduler for any compressor, guiding the selection of compression parameters. We present a Lightweight Efficient GrAdient Compression strategyY or LEGACY, which, in theory, can work with any compression technique to produce a simple dynamic counterpart. We benchmark LEGACY on distributed and federated training, involving seven different DNN architectures, ranging from ResNet, Transformer-XL, to GPT-2, across large and challenging datasets, including ImageNet, WikiText-103, and OpenWebText. On ImageNet-1K, with an equivalent average data volume, LEGACY's dynamic compression strategies improve the Top-1 accuracy of ResNet-50 by 7-11% compared to uniform Top-0.1% compression, while on WikiText-103, the layer-based dynamic strategy reduces the perplexity of Transformer-XL by ~26% relative to the same baseline. In addition, we evaluate LEGACY under constrained and federated settings, and demonstrate that it...},

keywords = {ICLR},

pubstate = {published},

tppubtype = {conference}

}

@conference{nokeyx,

title = {Pixel-Perfect Puppetry: Precision-Guided Enhancement for Face Image and Video Editing},

author = {Yan Li and Zhenyi Wang and Guanghao Li and Wei Xue and Yike Guo and Wenhan Luo

},

url = {https://openreview.net/forum?id=8mHZWTeF3z},

year  = {2026},

date = {2026-01-26},

urldate = {2026-01-26},

booktitle = {The Fourteenth International Conference on Learning Representations

},

abstract = {Preserving identity while precisely manipulating attributes is a central challenge in face editing for both images and videos. Existing methods often introduce visual artifacts or fail to maintain temporal consistency. We present FlowGuide, a unified framework that achieves fine-grained control over face editing in diffusion models. Our approach is founded on the local linearity of the UNet bottleneck’s latent space, which allows us to treat semantic attributes as corresponding to specific linear subspaces, providing a mathematically sound basis for disentanglement. FlowGuide first identifies a set of orthogonal basis vectors that span these semantic subspaces for both the original content and the target edit, a representation that efficiently captures the most salient features of each. We then introduce a novel guidance mechanism that quantifies the geometric alignment between these bases to dynamically steer the denoising trajectory at each step. This approach offers superior control by ensuring edits are confined to the desired attribute’s semantic axis while preserving orthogonal components related to identity. Extensive experiments demonstrate that FlowGuide achieves state-of-the-art performance, producing high-quality edits with superior identity preservation and temporal coherence.},

keywords = {ICLR},

pubstate = {published},

tppubtype = {conference}

}

@conference{nokeyy,

title = {Direct Preference Optimization for Primitive-Enabled Hierarchical RL: A Bilevel Approach},

author = {Utsav Singh and Souradip Chakraborty and Wesley Suttle and Brian Sadler and Derrik Asher and Anit Kumar Sahu and Mubarak Shah and Vinay Purushothaman Namboodiri and Amrit Bedi

},

url = {https://arxiv.org/html/2411.00361v3},

year  = {2026},

date = {2026-01-26},

booktitle = {Fourteenth International Conference on Learning Representations (ICLR), 2026.},

journal = {Fourteenth International Conference on Learning Representations (ICLR), 2026.},

abstract = {Hierarchical reinforcement learning (HRL) enables agents to solve complex, long-horizon tasks by decomposing them into manageable sub-tasks. However, HRL methods face two fundamental challenges: (i) non-stationarity caused by the evolving lower-level policy during training, which destabilizes higher-level learning, and (ii) the generation of infeasible subgoals that lower-level policies cannot achieve. To address these challenges, we introduce DIPPER, a novel HRL framework that formulates goal-conditioned HRL as a bi-level optimization problem and leverages direct preference optimization (DPO) to train the higher-level policy. By learning from preference comparisons over subgoal sequences rather than rewards that depend on the evolving lower-level policy, DIPPER mitigates the impact of non-stationarity on higher-level learning. To address infeasible subgoals, DIPPER incorporates lower-level value function regularization that encourages the higher-level policy to propose achievable subgoals. We introduce two novel metrics to quantitatively verify that DIPPER mitigates non-stationarity and infeasible subgoal generation issues in HRL. Empirical evaluation on challenging robotic navigation and manipulation benchmarks shows that DIPPER achieves upto 40% improvements over state-of-the-art baselines on challenging sparse-reward scenarios, highlighting the potential of preference-based learning for addressing longstanding HRL limitations.},

keywords = {ICLR},

pubstate = {published},

tppubtype = {conference}

}

@conference{nokeyz,

title = {TEST-TIME SCALING IN DIFFUSION LLMS VIA HIDDEN SEMI-AUTOREGRESSIVE EXPERTS},

author = {Jihoon Lee and Hoyeon Moon and Kevin Zhai and Arun Chithanar and Anit Kumar Sahu and Soummya Kar and Chul Lee and Souradip Chakraborty and Amrit Bedi

},

url = {https://openreview.net/forum?id=L5y7in91vd},

year  = {2026},

date = {2026-01-26},

urldate = {2026-01-26},

booktitle = {Fourteenth International Conference on Learning Representations (ICLR), 2026.



},

abstract = {Diffusion-based large language models (dLLMs) are trained to model extreme flexibility/dependence in the data-distribution; however, how to best utilize this at inference time remains an open problem. In this work, we uncover an interesting property of these models: dLLMs {trained on textual data} implicitly learn a mixture of semi-autoregressive experts, where different generation orders reveal different specialized behaviors. We show that committing to any single, fixed inference time schedule, a common practice, collapses performance by failing to leverage this latent ensemble. To address this, we introduce HEX (Hidden semi-autoregressive EXperts for test-time scaling), a training-free inference method that ensembles across heterogeneous block schedules. By doing a majority vote over diverse block-sized generation paths, HEX robustly avoids failure modes associated with any single fixed schedule. On reasoning benchmarks such as GSM8K, it boosts accuracy by up to 3.56× (from 24.72% to 88.10%), outperforming top-K margin inference and specialized fine-tuned methods like GRPO, without additional training. HEX even yields significant gains on MATH benchmark from 16.40% to 40.00%, scientific reasoning on ARC-C from 54.18% to 87.80%, and TruthfulQA from 28.36% to 57.46%. Our results establish test-time scaling as a powerful principle for dLLMs, showing that the sequence in which masking is done can play a significant role in test-time scaling/inferencing of dLLMs.},

keywords = {ICLR},

pubstate = {published},

tppubtype = {conference}

}

@conference{nokey_27,

title = {ATEX-CF: Attack-Informed Counterfactual Explanations for Graph Neural Networks},

author = {Yu Zhang and Bin Yang and ARIJIT KHAN and Cuneyt Akcora

},

url = {https://arxiv.org/abs/2602.06240},

year  = {2026},

date = {2026-01-26},

urldate = {2026-01-26},

booktitle = {Fourteenth International Conference on Learning Representations (ICLR), 2026.},

abstract = {Counterfactual explanations offer an intuitive way to interpret graph neural networks (GNNs) by identifying minimal changes that alter a model’s prediction, thereby answering “what must differ for a different outcome?”. In this work, we propose a novel framework, ATEX-CF that unifies adversarial attack techniques with counterfactual explanation generation—a connection made feasible by theirshared goal of flipping a node’s prediction, yet differing in perturbation strategy:adversarial attacks often rely on edge additions, while counterfactual methods typically use deletions. Unlike traditional approaches that treat explanation and attack separately, our method efficiently integrates both edge additions and deletions, grounded in theory, leveraging adversarial insights to explore impactful counterfactuals. In addition, by jointly optimizing fidelity, sparsity, and plausibility under a constrained perturbation budget, our method produces instance-level explanations that are both informative and realistic. Experiments on synthetic and real-world node classification benchmarks demonstrate that ATEX-CF generates faithful, concise, and plausible explanations, highlighting the effectiveness of integrating adversarial insights into counterfactual reasoning for GNNs.},

keywords = {ICLR},

pubstate = {published},

tppubtype = {conference}

}

@conference{nokey_28,

title = {TopoFormer: Topology Meets Attention for Graph Learning},

author = {Md Joshem Uddin and Astrit Tola and Cuneyt Akcora and Baris Coskunuzer

},

url = {https://openreview.net/forum?id=xanpQt9YZh},

year  = {2026},

date = {2026-01-26},

urldate = {2026-01-26},

booktitle = {Fourteenth International Conference on Learning Representations (ICLR), 2026.



},

abstract = {We introduce TopoFormer, a lightweight and scalable framework for graph representation learning that encodes topological structure into attention-friendly sequences. At the core of our method is Topo-Scan, a novel module that decomposes a graph into a short, ordered sequence of topological tokens by slicing over node or edge filtrations. These sequences capture multi-scale structural patterns, from local motifs to global organization, and are processed by a Transformer to produce expressive graph-level embeddings. Unlike traditional persistent homology pipelines, Topo-Scan is parallelizable, avoids costly diagram computations, and integrates seamlessly with standard deep learning architectures. We provide theoretical guarantees on the stability of our topological encodings and demonstrate state-of-the-art performance across graph classification and molecular property prediction benchmarks. Our results show that TopoFormer matches or exceeds strong GNN and topology-based baselines while offering predictable and efficient compute. This work opens a new path for parallelizable and unifying approaches to graph representation learning that integrate topological inductive biases into attention frameworks.},

keywords = {ICLR},

pubstate = {published},

tppubtype = {conference}

}

@conference{nokey_29,

title = {Privacy Beyond Pixels: Latent Anonymization for Privacy-Preserving Video Understanding},

author = {Joseph Fioresi and Ishan Rajendrakumar Dave and Mubarak Shah

},

url = {https://arxiv.org/abs/2511.08666},

year  = {2026},

date = {2026-01-26},

booktitle = {Fourteenth International Conference on Learning Representations (ICLR), 2026.



},

abstract = {We introduce a novel formulation of visual privacy preservation for video foundation models that operates entirely in the latent space. While spatio-temporal features learned by foundation models have deepened general understanding of video content, sharing or storing these extracted visual features for downstream tasks inadvertently reveals sensitive personal information like skin color, gender, or clothing. Current privacy preservation methods focus on input-pixel level anonymization, which requires retraining the entire utility video model and results in task-specific anonymization, making them unsuitable for recent video foundational models. To address these challenges, we introduce a lightweight Anonymizing Adapter Module (AAM) that removes private information from video features while retaining general task utility. AAM can be applied in a plug and play fashion to frozen video encoders, minimizing the computational burden of finetuning and re-extracting features. Our framework employs three newly designed training objectives: (1) a clip-level self-supervised privacy objective to reduce mutual information between static clips, (2) a co-training objective to retain utility across seen tasks, and (3) a latent consistency loss for generalization on unseen tasks. Our extensive evaluations demonstrate a significant 35% reduction in privacy leakage while maintaining near-baseline utility performance across various downstream tasks: Action Recognition (Kinetics400, UCF101, HMDB51), Temporal Action Detection (THUMOS14), and Anomaly Detection (UCF-Crime). We also provide an....},

keywords = {ICLR},

pubstate = {published},

tppubtype = {conference}

}

@conference{nokey_30,

title = {Agent-X: Evaluating Deep Multimodal Reasoning in Vision-Centric Agentic Tasks},

author = {Tajamul Ashraf and Amal Saqib and Hanan Gani and Muhra AlMahri and Yuhao Li and Noor Ahsan and Umair Nawaz and Jean Lahoud and Hisham Cholakkal and Mubarak Shah and Philip Torr and Fahad Khan and Rao Anwer and Salman Khan},

url = {https://openreview.net/pdf?id=Vjruxvp1Xd},

year  = {2026},

date = {2026-01-26},

urldate = {2026-01-26},

booktitle = {Fourteenth International Conference on Learning Representations (ICLR), 2026.



},

abstract = {Deep reasoning is fundamental for solving complex tasks, especially in vision-centric scenarios that demand sequential, multimodal understanding. However, existing benchmarks typically evaluate agents with fully synthetic, single-turn queries, limited visual modalities, and lack a framework to assess reasoning quality over multiple steps as required in real-world settings. To address this, we introduce Agent-X, a large-scale benchmark for evaluating vision-centric agents’ multistep and deep reasoning capabilities in real-world, multimodal settings. AgentX features 828 agentic tasks with authentic visual contexts, including images, multi-image comparisons, videos, and instructional text. These tasks span six major agentic environments: general visual reasoning, web browsing, security and surveillance, autonomous driving, sports, and math reasoning. Our benchmark requires agents to integrate tool use with explicit, stepwise decision-making in these diverse settings. In addition, we propose a fine-grained, step-level evaluation framework that assesses the correctness and logical coherence of each reasoning step and the effectiveness of tool usage throughout the task. Our results reveal that even the best-performing models, including GPT, Gemini, and Qwen families, struggle to solve multi-step vision tasks, achieving less than 50% full-chain success. These findings highlight key bottlenecks in current LMM reasoning and tool-use capabilities and identify future research directions in vision-centric agentic reasoning models},

keywords = {ICLR},

pubstate = {published},

tppubtype = {conference}

}

@conference{nokey_31,

title = {Sample-Efficient Distributionally Robust Multi-Agent Reinforcement Learning via Online Interaction},

author = {Zain Ulabedeen Farhat and Debamita Ghosh and George Atia and Yue Wang

},

url = {https://openreview.net/forum?id=hIBMCDOwMI},

year  = {2026},

date = {2026-01-26},

urldate = {2026-01-26},

booktitle = {Fourteenth International Conference on Learning Representations (ICLR), 2026.



},

abstract = {Well-trained multi-agent systems can fail when deployed in real-world environments due to model mismatches between the training and deployment environments, caused by environment uncertainties including noise or adversarial attacks. Distributionally Robust Markov Games (DRMGs) enhance system resilience by optimizing for worst-case performance over a defined set of environmental uncertainties. However, current methods are limited by their dependence on simulators or large offline datasets, which are often unavailable. This paper pioneers the study of online learning in DRMGs, where agents learn directly from environmental interactions without prior data. We introduce the Multiplayer Optimistic Robust Nash Value Iteration (MORNAVI) algorithm and provide the first provable guarantees for this setting. Our theoretical analysis demonstrates that the algorithm achieves low regret and efficiently finds the optimal robust policy for uncertainty sets measured by Total Variation divergence and Kullback-Leibler divergence. These results establish a new, practical path toward developing truly robust multi-agent systems.},

keywords = {ICLR},

pubstate = {published},

tppubtype = {conference}

}

@conference{nokeyo,

title = {Learning from Noisy Preferences: A Semi-Supervised Learning Approach to Direct Preference Optimization},

author = {Xinxin Liu and Ming Li and Zonglin Lyu and Yuzhang Shang and Chen Chen},

url = {https://openreview.net/forum?id=rRc04jyoAk},

year  = {2026},

date = {2026-01-26},

urldate = {2026-01-26},

booktitle = {The Fourteenth International Conference on Learning Representations},

abstract = {Human visual preferences are inherently multi-dimensional, encompassing aspects of aesthetics, detail fidelity, and semantic alignment. However, existing open-source preference datasets provide only single, holistic annotations, resulting in severe label noise—images that excel in some dimensions (e.g., compositional) but are deficient in others (e.g., details) are simply marked as winner" orloser". We theoretically demonstrate that this compression of multi-dimensional preferences into binary labels generates conflicting gradient signals that misguide the optimization process in Diffusion Direct Preference Optimization (DPO). To address this label noise from conflicting multi-dimensional preferences, we propose Semi-DPO, a semi-supervised learning approach. We treat pairs with consistent preferences across all dimensions as clean labeled data, while those with conflicting signals are considered noisy unlabeled data. Our method first trains a model on a clean, consensus-filtered data subset. This model then acts as its own implicit classifier to generate pseudo-labels for the larger, noisy set, which are used to iteratively refine the model's alignment. This approach effectively mitigates label noise and enhances image generation quality, achieving better alignment with multi-dimensional human preferences. Experimental results demonstrate that Semi-DPO significantly improves alignment with multi-dimensional human preferences, achieving state-of-the-art performance without requiring additional human annotation or the need to train a dedicated reward models.},

keywords = {ICLR},

pubstate = {published},

tppubtype = {conference}

}