Leveraging foundation vision models for cross-species animal pose tracking, achieving high accuracy with limited labels.

Le Li, Daniela Ivanova, Nicolas Pugeault

Marginal matching does not ensure factorized sampling; model leakage persists.

Duong Bach, Hai Nguyen Hong, Cuong Do

Introduces variational bounds for the finite-temperature perceptron trained on Gaussian mixtures, deriving upper and lower bounds via interpolation and log-concavity, with fixed point equations for energy and error estimates.

Francesco Camilli, Pierluigi Contucci, Federica Gerace et al.

MemoryCPT employs an end-to-end trainable memory framework combining QAD and QAR, achieving a QPC of 0.138 with significant cost reduction and answer quality improvement.

Songxin Lei, Kun Ouyang, Weilin Ruan et al.

OCSD compares matched replay views to calibrate environment feedback, significantly improving multi-task reinforcement learning performance.

Yi Yang, Cong Qin, Xiaodan Liu et al.

Proposes TriQua framework, combining base triples with qualifiers for fine-grained factuality evaluation, achieving high correlation (r=0.89) with human scores.

Jin Liu, Steffen Thoma, Achim Rettinger

MAGIC combines graph label propagation and geometric alignment to stabilize feature space in semi-supervised class-incremental learning, reducing drift and improving accuracy.

Yousef Abdi, Mohammad Asadpour, Yousef Seyfari

Masked Diffusion model improves beat tracking by modeling multiple plausible outputs, reducing incoherence and enhancing stability.

Francesco Foscarin, Filip Korzeniowski, Richard Vogl

ABD method achieves 59.43% accuracy on LiveCodeBench-v6, outperforming Single9 and HAC.

Ruitong Li, Binjie Guo, Aisheng Mo et al.

Sun develops the ISMIE framework, combining crowdsourcing surveys, neurophysiological signals, and preference modeling to analyze GenAI's impact on modern information seeking, revealing preference shifts and cognitive load dynamics.

Shuoqi Sun

Otter is a 15.3M-parameter chess AI using history and time signals to predict human moves with 55.23% accuracy.

Tarun Kumar S

EuroExec benchmark reveals frontier LLMs achieve only 56.9% solve rate on European executive tasks, far below expert performance.

Pau Arnal, Khaled Denfir, Danylo Smahliuk et al.

Introduces DiVers dataset with over 1.1 million music versions, enhancing robustness in real-world scenarios, significantly improving music version identification (VI) performance.

Simon Hachmeier, R. Oguz Araz, Dmitry Bogdanov et al.

Proposes a multi-objective ranking framework combining immediate and delayed signals with Segment-Aware targeting, boosting Twitch DAU by 0.09% and ARPU by 0.56%, with 41.9% fewer parameters.

Xiaoyi Gu, Julia Tavares, Eder Santana et al.

CommBench benchmarks LLMs for GPU communication code correctness and efficiency; GPT-5.5 achieves only 30.7% success on tasks.

Shuang Ma, Yuyi Li, Yihan Zhang et al.

ToolArtist employs post-trained UMM with RL and RAD-GRPO to dynamically coordinate reasoning, tool use, and image generation, outperforming fixed pipeline methods.

Jiahao Zhao, Xiaomin Yu, Zhongxiang Sun et al.

BinaryPC employs data-aware binary principal components for training-free sparse attention, maintaining accuracy and boosting decoding throughput by 3.56×.

Daohai Yu, Zhanpeng Zeng, Keyu Chen et al.

iStructTab optimizes multimodal learning of images and tabular data using GEDS, enhancing predictive performance.

Al Zadid Sultan Bin Habib, Md Younus Ahamed, Prashnna Gyawali et al.

Introduces CLIP-CC-Bench, a multi-model ensemble framework for evaluating paragraph-level video descriptions, covering 17 SOTA models with 200 movie clips.

Mukhtiar Ali, Harsh Dubey, Sugam Mishra et al.

A spanning-tree overlap bound certifies frozen LLM pipelines; 12 configurations achieved 92.7%±2.4% trajectory coverage.

Zhenpeng Li