The paper investigates the impact of low‑resolution data on machine learning models, especially when high‑resolution data is scarce. It provides a theoretical analysis using Kullback‑Leibler divergence to quantify how data resolution affects a datapoint’s influence and derives bounds on information loss from downsampling. Empirical results with a vision transformer and a convolutional neural network show that incorporating low‑resolution data consistently improves performance in such settings.
By Theresa Dahl Frehr, Niels Henrik Pontoppidan, Hiba Nassar, Tommy Sonne Alstr{\o}m
The paper investigates how alignment—both outer (choosing which graphs to pair) and inner (aligning node representatives)—affects permutation-equivariant graph flow matching. It connects inner alignment to transport on the graph quotient space and shows that quotient couplings can be lifted to aligned representatives, while symmetrization yields equivariant flow‑matching minimizers. Experiments on continuous graph and molecular generation demonstrate that appropriate alignment can simplify trajectories and improve few‑step generation, though the benefits vary with the type of alignment and computational budget.
By Moritz Piening, Christian Wald
arXiv:2609.34373v2 Announce Type: replace
Abstract: Rate-limited multi-agent teams raise three questions the emergent-communication literature has answered only empirically: what an optimal message s...
By Mihir Chauhan, Aniket Bera
arXiv:2609.34426v2 Announce Type: replace
Abstract: This paper addresses the problem of safe offline reinforcement learning, which involves training a policy to satisfy safety constraints using an of...
By Shengchao Hu, Peng Wang, Jifeng Hu, Qiyang Zhou, Anning Hu, Li Shen, Ya Zhang, Dacheng Tao
The paper introduces Alignment‑Guided Flow Transformer (AGFT), a framework for Vision‑Language‑Action (VLA) models that explicitly enforces tri‑modal alignment among vision, language, and action through a dedicated alignment loss. AGFT bridges representational gaps across modalities, improving task adaptation and robustness, and employs a flow‑matching objective to reduce inference steps compared to diffusion‑based policies. Experiments on a large benchmark demonstrate that AGFT achieves higher success rates and lower inference latency than state‑of‑the‑art baselines, highlighting tri‑modal alignment as crucial for scalable VLA manipulation.
By Shengchao Hu, Peng Wang, Qiyang Zhou, Guodong Zheng, Yuqi Huang, Li Shen, Ya Zhang, Dacheng Tao
arXiv:2501.12632v3 Announce Type: replace-cross
Abstract: Weakly supervised object localization (WSOL) models can predict both the object class and the spatial regions corresponding to the object, wi...
By Shakeeb Murtaza, Soufiane Belharbi, Alexis Guichemerre, Marco Pedersoli, Eric Granger
arXiv:2505.20817v3 Announce Type: replace-cross
Abstract: Gradient clipping is widely used in language-model training to control heavy-tailed gradient noise and can improve convergence guarantees ove...
By Taha El Bakkali El Kadi, Savelii Chezhegov, Aleksandr Beznosikov, Samuel Horv\'ath, Eduard Gorbunov
The paper introduces AttWarp, a lightweight technique that uses a multimodal large language model’s cross‑modal attention to perform rectilinear warping of input images at test time. By reallocating spatial resolution toward query‑relevant regions without altering model weights or architecture, AttWarp preserves global context while making small objects and subtle relationships easier for the model to read. Experiments on five benchmarks and four MLLMs show consistent accuracy gains, improved compositional reasoning, and reduced hallucinations compared to baseline image‑manipulation methods.
By Dwip Dalal, Gautam Vashishtha, Utkarsh Mishra, Jeonghwan Kim, Madhav Kanda, Hyeonjeong Ha, Svetlana Lazebnik, Heng Ji, Unnat Jain
The paper investigates Partial Least Squares (PLS) in high-dimensional settings, focusing on a model where two data matrices share a low-rank latent structure plus individual-specific components. By analyzing the singular vectors of the cross‑covariance matrix with random matrix theory, the authors derive asymptotic characterizations of how well the estimated latent directions align with the true ones. They show that the PLS variant based on Singular Value Decomposition (PLS‑SVD) outperforms separate principal component analysis in detecting the common latent subspace, while also identifying regimes where PLS‑SVD behaves counter‑intuitively or reaches fundamental limits.
By Victor L\'eger, Florent Chatelain
arXiv:2603.28773v2 Announce Type: replace-cross
Abstract: Large language models (LLMs) frequently generate confident yet factually incorrect content when used for language generation (a phenomenon of...
By Dobrik Georgiev, Kheeran K. Naidu, Alberto Cattaneo, Federico Monti, Carlo Luschi, Daniel Justus
The paper introduces CAM-Steer, a Category‑Adaptive Multi‑category Safety Steering framework that estimates risk for each harm category by comparing hidden states to safe and unsafe prototypes. It then combines safety directions into a single steering vector and applies a rotation whose angle is set by the estimated risks, preserving the hidden‑state norm. Experiments on three LLM backbones and seven harm categories show that CAM‑Steer outperforms baselines in defense success rate, even when multiple harm categories co‑occur, with negligible inference overhead.
By Chenxi Wang, Ruiyang Huang, Li Huang, Yifan Wu
arXiv:2609.35232v2 Announce Type: replace-cross
Abstract: Visual-token compression is effective for improving the efficiency of vision-language models, but under extreme compression budgets, token pr...
By Rui Zhong, Yu Li, Zheyu Yan, Cheng Zhuo
The paper introduces Event‑Grounded Self‑Distillation (EGSD), a method for real‑time video understanding that treats streaming memory as incremental updates over verifiable events such as key visual entities, actions, and details. EGSD adapts on‑policy self‑distillation by combining a multiplicative weight with outcome reward to align teacher and student preferences, and re‑weights the teacher with event information plus an entity‑coverage reward to counter question‑relevance bias. Experiments on StreamingBench and OVO‑Bench Real‑Time track show EGSD achieves 79.8 % and 73.4 % accuracy respectively, while improving effective‑entity recall by 17.4 % with only a 6.8 % increase in memory length.
By Yuwei Miao, Xuesheng Zhang, Wenhao Zou, Jixia Zhang, Jianwei Lv, Bo Yuan, Junfeng Wang, Shiao Xie
Waypoint‑1.5 is a real‑time diffusion world model designed for interactive video generation on consumer‑grade hardware. It is pre‑trained on 100,000 hours of control‑aligned video game footage and can generate playable video conditioned on full keyboard and mouse input. The system offers two resolution variants, distinguishes rendered FPS, latent FPS, and control rate, and includes a detailed data pipeline, architecture, training methodology, and runtime system.
By Rajit Rajpal, Shahbuland Matiana, Liew Wei Pyn, Anmol Agarwal, Ryan Craig, Andrew Lapp, Mithun Hunsur, Sami BuGhanem, Scottie Fox, Aaron Sanders Carson Poole, Irene Park, Dave Rossi, Spencer Frazier, Louis Castricato
arXiv:2609.37098v1 Announce Type: cross
Abstract: Vehicle-infrastructure cooperation can complement onboard sensing with broader and more informative observations of the traffic environment, providin...
By Junwei You, Weizhe Tang, Can Wang, Yan Zhao, Jun Hua, Haotian Shi, Wei Zhang, Lin Wang, Bin Ran
arXiv:2609.37786v1 Announce Type: new
Abstract: Concept Bottleneck Models (CBMs) built on vision-language models such as CLIP represent a latent space as human-understandable concepts. These represen...
By R\'emi Kazmierczak, Johanne Cohen, Marianne Clausel
SafeCoEvo is a test‑time framework that co‑evolves safety harnesses and guards for large language model agents. It uses a short‑term S‑Harness to quickly externalize recent runtime experience into explicit safety knowledge, and a long‑term GuardVPO to internalize accumulated experience into parametric risk‑judgment capabilities. This dual adaptation improves safety and task success, reducing unsafe outcomes by 10.05% and increasing task success by 12.15% over the strongest baseline.
By Yu Cheng, Yongkang Hu, Shuaijie Ma, Zhihang Lin, Weicheng Meng, Jingyang Qiao, Jiuan Zhou, Yushuo Zhang, Yihang Chen, Weilin Luo, Kun Shao, Dong Li, Zhizhong Zhang, Yuan Xie, Zhaoxia Yin
The paper introduces BRAID, a hierarchical latent-variable model that generates multi-person human motion by explicitly modeling both group-level interaction dynamics and individual behavior conditioned on evolving group context. It treats social motion generation as a meta-transfer learning problem, learning shared interaction priors across datasets and adapting them to arbitrary context sets of observed people and joints. BRAID supports coherent generation under full, sparse, or partial observations and produces compact social-state vectors useful for downstream embodied-agent systems, with evaluations on social forecasting, tracking, in-filling, and response generation.
By Ojas Shirekar, Yash Surange, Agustinas Ju\v{c}as, Chirag Raman
arXiv:2609.36956v1 Announce Type: cross
Abstract: Manipulating next-token probabilities during generation can bypass the safety alignment of large language models. Existing approaches, however, rely...
By Jesson Wang, Shawn Li, Wei Yang, Franck Dernoncourt, Ryan A. Rossi, Charith Peris, Yue Zhao
Direct-decision models turn text into low-latency structured labels and scores, making them attractive for classification and automatic evaluation. Yet reliability requires more than accuracy: a model...