CoARS framework enhances recommender systems using self-distilled reinforcement learning, showing improved user alignment in experiments.

Zongwei Wang, Min Gao, Hongzhi Yin et al.

Study shows vision transformers rely on Gestalt continuity for object binding, with probes sensitive to continuity across models.

Alexa R. Tartaglini, Michael A. Lepori

K-STEMIT integrates geometric, temporal, and physical priors in a multi-branch GNN for ice layer thickness estimation, reducing RMSE by 21%.

Zesheng Liu, Maryam Rahnemoonfar

MAGE framework enables conditional music generation and mixture-grounded target-source extraction in a shared continuous latent space.

Muhammad Usama Saleem, Tejasvi Ravi, Tianyu Xu et al.

Proposes GIANTS, a reinforcement learning-based model that improves scientific insight prediction by 34% on 17k samples, outperforming proprietary baselines.

Joy He-Yueya, Anikait Singh, Ge Gao et al.

RecaLLM addresses the 'lost-in-thought' issue with explicit in-context retrieval, excelling on RULER and HELMET benchmarks.

Kyle Whitecross, Negin Rahimi

Kernel-based RKHS method for Koopman eigenfunctions, preserving spectrum for high-dimensional stability analysis.

François-Grégoire Bierwart, Alexandre Mauroy

GREATEN integrates surface normals with efficient sparse attention to enhance cross-domain stereo matching, reducing errors by over 30%.

Jiahao Li, Xinhong Chen, Zhengmin Jiang et al.

VLA-World integrates predictive imagination with reflective reasoning, reducing collision rate to 0.10% and outperforming SOTA in autonomous driving benchmarks.

Guoqing Wang, Pin Tang, Xiangxuan Ren et al.

Watt Counts optimizes LLM energy efficiency on heterogeneous GPUs using a dataset of 5,000+ experiments, achieving up to 70% energy savings.

Mauricio Fadel Argerich, Jonathan Fürst, Marta Patiño-Martínez

Proposed Neighborhood Transformer (NT) with switchable attention outperforms SOTA in node classification, especially on heterophilic graphs, reducing complexity significantly.

Yi Luo, Xu Sun, Guangchun Luo et al.

MedConceal uses an interactive patient simulator to evaluate hidden-concern reasoning in medical dialogue, featuring 300 cases and 600 interactions.

Yikun Han, Joey Chan, Jingyuan Chen et al.

Hierarchical VLM-guided semantic-geometric framework improves task-aware bimanual grasping, achieving 88.9% strategy alignment in real-world tests.

Fabian Hahne, Vignesh Prasad, Georgia Chalvatzaki et al.

Sumo method enables quadruped robots to dynamically manipulate through sample-based planning, solving multi-task problems.

John Z. Zhang, Maks Sorokin, Jan Brüdigam et al.

SkillClaw turns multi-user experience into shared skill evolution; on WildClawBench, Qwen3-Max gains up to 52% in Search & Retrieval.

Ziyu Ma, Shidong Yang, Yuxiang Ji et al.

We propose a feedforward graph architecture using frozen large language models as nodes, achieving 87.3% accuracy on ARC-Challenge.

Marcus Armstrong, Navid Ayoobi, Arjun Mukherjee

Analyzes the impact of objective misspecifications on stability and sensitivity in model predictive game controllers.

Ada Yildirim, Bryce L. Ferguson

OV-Stitcher employs a global attention mechanism without training, boosting mIoU from 48.7 to 50.7 by stitching sub-image features for open-vocabulary segmentation.

Seungjae Moon, Seunghyun Oh, Youngmin Ro

Bridge-STG improves video grounding by decoupling spatio-temporal alignment, boosting m_vIoU from 26.4 to 34.3.

Xuezhen Tu, Jingyu Wu, Fangyu Kang et al.

By decomposing long-context reasoning into atomic skills, the method enhances LLMs' reasoning by an average of 7.7%.

Yanling Xiao, Huaibing Xie, Guoliang Zhao et al.