The paper introduces a Bayesian non‑parametric method for clustering time series that allows an unbounded number of clusters to evolve by switching among an unknown number of regimes with linear dynamics. It employs a hierarchical Dirichlet process prior for the Switching Linear Dynamical System parameters and a Gaussian process prior to capture amplitude and temporal alignment variations within each cluster. The approach formulates a variational lower bound for both offline and online inference, enabling efficient learning, and demonstrates its effectiveness on electrocardiogram datasets.
By Adri\'an P\'erez-Herrero, Paulo F\'elix, Jes\'us Presedo, Carl Henrik Ek
Ensemble-Aware Structural Encoding (EASE) is a new framework that conditions protein encoders on peptide conformational ensembles rather than single sequences or structures. By selecting central conformers that capture a peptide’s major structural states and fusing their representations with a permutation‑invariant operator, EASE produces a peptide‑level representation that better reflects conformational variability. In evaluations on six peptide property‑prediction tasks, EASE consistently outperforms single‑structure baselines, achieving an average absolute gain of about 3.0% and improving further as more conformers are incorporated.
By Ruoyao Zhang, Hongxuan Li, Jie Peng, Tianlong Chen
arXiv:2610.27653v1 Announce Type: cross
Abstract: Watermarking has emerged as a key tool for identifying text generated by large language models (LLMs), with current schemes embedding statistical sig...
By Romina Omidi, Nima Naderloui, Yuan Hong, Binghui Wang
The paper shows that in text‑to‑image evaluation, human preference for attractive faces can outweigh objective image quality, leading to biased rankings and model outputs. A user study revealed that annotators consistently rate more attractive faces as higher quality, even when comparing real photographs of equal fidelity. This bias can invert leaderboard rankings, dominate preference model scores, and cause generated faces to shift toward younger, more feminine features, undermining user control.
By Harvey Mannering, Zhiwu Huang, Adam Pr\"ugel-Bennett
arXiv:2610. 25116v2 Announce Type: replace Abstract: We study positive, bias-free, one-hidden-layer ReLU networks under standard Gaussian inputs.
By Zhijie Zheng
The paper introduces Adaptive Dual-Teacher Distillation with Subnetwork Rectification (DDSR), a method for black-box domain adaptation that fuses predictions from a black-box source model and a vision‑language model to generate reliable pseudo‑labels for the target domain. DDSR uses adaptive prediction fusion, subnetwork regularization to prevent overfitting, iterative refinement of pseudo‑labels and prompts, and class‑wise prototypes for self‑training. Experiments on several benchmark datasets show that DDSR outperforms existing state‑of‑the‑art approaches, even those that have access to source data or source model parameters.
By Zhe Zhang, Jing Li, Wanli Xue, Xu Cheng, Jianhua Zhang, Qinghua Hu, Shengyong Chen
The paper explores activation steering as a method to enhance automatic speech recognition (ASR) performance and demographic fairness. It compares two techniques—paired activation subtraction and supervised optimization—finding that supervised optimization effectively improves overall ASR accuracy and fairness across gender, age, and dialect. The study also demonstrates that steering vectors trained on acoustically similar speakers can yield competitive or superior results, enabling few‑shot or zero‑shot adaptation without requiring demographic labels.
By Felix Herron, Alex Colagrande, Solange Rossato, Fran\c{c}ois Portet, Alexandre Allauzen, Chuyuan Li
The paper introduces SHAM-Bench, a unified benchmark for detecting span-level hallucinations across six natural language generation tasks, compiled from 6,461 human-annotated examples under a common schema. It evaluates existing detection methods, finding that fine‑tuned detectors perform well in‑domain but falter under distribution shift, while prompt‑based methods with reasoning show greater robustness. The study also highlights challenges such as subtle unsupported content, ambiguous span boundaries, and noisy labels, and releases the dataset, evaluation code, and a public leaderboard for future research.
By Sicong Huang, Ian Lane
The paper investigates whether linguistic and arithmetic tasks in large language models (LLMs) rely on shared neural circuits. Using mechanistic interpretability techniques on five LLMs, the authors find that circuit similarity aligns primarily with task domain rather than input/output vocabularies or syntactic structure. This suggests that LLMs may employ largely separate neural circuits for language and arithmetic, indicating a modular organization of abilities.
By Riku Kisako, Tatsuki Kuribayashi, Ryohei Sasano
The paper investigates how reward misspecification in reinforcement learning harms the ability to monitor chain-of-thought (CoT) reasoning in language models. By creating a testbed with true and misspecified rewards, the authors show that models can hack the reward while still producing CoTs that appear aligned, leading to low monitorability. They propose a mitigation strategy that collects diverse reward rollouts, infers a reward description with a language model, and injects this description into the input, improving monitorability by up to 42% absolute across models and tasks.
By Tianyi Lorena Yan, John Hewitt
The paper introduces RoboJudge, a multimodal language model trained to evaluate embodied video generation. It presents RoboJudge-Bench, a benchmark of 800 human‑annotated videos assessing Physical Adherence and Instruction Alignment, and RoboJudge-Train, a 12K‑video dataset with human‑verified scores and rationales. Fine‑tuned on this data, RoboJudge‑9B achieves the highest agreement with human judgments, outperforming existing models and improving performance on external physics benchmarks.
By Siyuan Ma, Yu Qi, Cheng Ye, Ruotian Peng, Xinyi Xu, Zitiantao Lin, Zedong Cai, Yijian Huang, Weiyang Liu, Siyuan Wang, Yandong Wen
Latent Bridge connects video generative latent spaces with geometric feature spaces, enabling direct prediction of 3D geometry—cameras, depth, and point clouds—without an intermediate RGB decoding step. The method trains representation‑specific bridges for Wan2.1 and V‑RAE via feature distillation on RGB videos while keeping the video encoder and geometry model frozen, requiring no ground‑truth geometry. At inference, the same video latents support both RGB decoding and geometry prediction, and the trained bridge can be reused across image‑to‑video and text‑to‑video generation, achieving competitive reconstruction quality on Co3Dv2 and WildRGB‑D with efficient end‑to‑end generation.
By Shuai Zhang, Zhanpeng Ouyang, Zhengqing Chen, Ruohong Mei, Shaoyu Chen, Weiqiang Ren, Hui Sun, Qian Zhang, Wenyu Liu, Xinggang Wang
The paper investigates the theoretical relationship between maximum clique and maximum consensus formulations used in robust point cloud registration. It demonstrates that a maximum clique can incur arbitrarily large alignment error and be much larger than the maximum consensus set at any fixed tolerance, while also deriving geometric conditions that bound this error. The authors provide sharp error bounds for the translation subproblem with fixed rotation, analyze group‑k consistency, and identify the minimal group sizes needed to guarantee alignment error bounds independent of scale and geometry.
By Abolfazl Babanazari, Kaveh Fathian
TrajTok introduces an end‑to‑end video tokenizer that replaces traditional patchification with a lightweight, trajectory‑based tokenization process. It performs implicit clustering over space‑time pixels to generate object trajectories in a single forward pass, dynamically adjusting token granularity to semantic complexity while remaining independent of video duration. The resulting TrajViT2 model outperforms existing methods on classification and retrieval benchmarks, and TrajTok can also serve as a probing head (TrajAdapter) or alignment connector (TrajVLM) for vision‑language tasks, notably improving long‑video reasoning.
By Chenhao Zheng, Jieyu Zhang, Jianing Zhang, Weikai Huang, Ashutosh Kumar, Quan Kong, Oncel Tuzel, Chun-Liang Li, Ranjay Krishna
The paper introduces a plug‑in estimator for the non‑iterative conditional expectation (NICE) g‑formula that uses long short‑term memory (LSTM) models instead of traditional parametric models. It demonstrates, through a complex data‑generation scenario resembling an HIV cohort, that the LSTM‑based estimator yields lower bias and near‑nominal 95% confidence‑interval coverage compared to parametric alternatives. The authors suggest this work as a step toward a theoretical framework for deep‑learning plug‑in estimators in high‑dimensional longitudinal causal inference.
By Jing Li, Sophia M. Rein, Zilong Wang, Zixiao Wang, Sean McGrath, James M. Robins, Thomas Kolokotrones, Rajarshi Mukherjee, Miguel A. Hern\'an
arXiv:2610.24264v2 Announce Type: replace
Abstract: As LLM-based agents are deployed with increasing autonomy in real-world environments, a critical failure mode emerges: agentic misalignment, where...
By Eric Zhang, Terry Jingchen Zhang, Chijioke Ugwuanyi, Jerick Shi, Bernhard Sch\"olkopf, Zhijing Jin
arXiv:2610.27691v1 Announce Type: new
Abstract: Modern industrial ranking systems increasingly adopt generative recommendation (GR) paradigms, in which user behavior sequences and non-sequential (NS)...
By Bohao Li, Dinglun Cao, Pengcheng Huang, Yingda Tang, Zhiting Zhao, Yi Cao, Jilei Fu, Miao Yan, Tianying Liu, Heng Chang, Bowen Du
arXiv:2610. 26964v1 Announce Type: new Abstract: Large language models (LLMs) struggle with classical planning, but benchmarks rarely distinguish structural difficulty from the difficulty of interpreting symbolic vocabulary.
By Mayukh Debnath, Adeyinka Adedigba, Rammohan Mallipeddi
The paper examines inference‑time observation attacks on Safe Multi‑Agent Reinforcement Learning (SMARL), where an adversary perturbs observations of some agents to alter their actions. It introduces the MinMax Attack framework, which jointly selects adversarial target actions for compromised agents while considering the cooperative resistance of benign agents. The authors derive a bound on the induced safety cost and show through experiments that MinMax Attack causes higher safety costs than existing baselines while maintaining task reward, thus improving attack effectiveness and stealthiness.
By Taozheng Zhu, Shixiong Jiang, Fanxin Kong
arXiv:2610.25964v2 Announce Type: replace-cross
Abstract: We set out to reproduce a recent hallucination-mitigation result on the POPE benchmark. The LLaVA-1.5-7B baseline we measured, 82.21 F1 on th...
By Kesav Kumar Jayakumar, Karthigeyan Thilak