Introduces covariance neural networks (VNNs), combining PCA and graph learning to improve stability and transferability on multiscale datasets.

Saurabh Sihag, Andrea Cavallo, Elvin Isufi et al.

Introduced Semigroup-JEPA for zero-shot physics generalization, achieving 34% error reduction and 23.3% control success improvement.

Andy Zeyi Liu, Haoran Sun, Lucas Baker et al.

JarvisGUI evaluates cross-device GUI agents with dynamic task composition, revealing capability gaps in real-world workflows.

Zixiang Chen, Yuheng Lu, Zihao Cheng et al.

ConvMem reformulates long-context reasoning as hierarchical convolution, outperforming training-free baselines on RULER-HotpotQA.

Hongming Zhang, Zhaozhen Gu, Fengshuo Bai et al.

FOM-UL achieves efficient forgetting by selectively updating Transformer layers, maintaining utility and privacy under 8/4-bit quantization.

Ravi Ranjan, Olivera Kotevska, Agoritsa Polyzou

Fortunate Recall uses a 10+1 behavioral ontology and lifecycle policies to enhance LLM memory management, achieving 76.9% on LifecycleBench.

Ansuman Mullick, Eray Tüzün

FreqFM improves VLA models by frequency conditioning, achieving a 9.3-point gain on LIBERO-Plus.

Haochen Niu, Shengye Dong, Hao Liu et al.

MOONWALK optimizes animation/VFX pre-production reviews with an intent-evidence-action alignment framework, improving intent clarity and task executability.

Shih-Yu Lai, Wen-Fan Wang, Sai Ling et al.

DRiF framework uses data-driven risk fields for safer end-to-end driving, achieving 88.78 driving score on Bench2Drive.

Yuanxin Tian, Zhiyuan Liu, Jinhao Li et al.

PACE reduces P95 PTFR from 0.53s to 0.29s via perceived-latency-aware dialogue routing.

Lin Huang, Yujuan Tan, Weisheng Li et al.

Survey of inference-efficiency mechanisms in VideoLLMs, analyzing frame sampling, modality encoding, and token compression techniques.

Killian Steunou, Yannis Tevissen, Mounîm A. El Yacoubi

Proposed CAMF achieves 91.86% intention recognition accuracy for industrial human-robot collaboration.

Xinyu Liu, Qiqi Dong, Boya Jia et al.

Motar achieves 15.4 FPS high-fidelity streaming talking head generation by fusing conditions in a low-dimensional motion space.

Yanru An, Ruiyan Wang, Wenwu Wei et al.

ScopeMamba-YOLO enhances small object detection in remote sensing imagery by widening perceptual scope, achieving a 10.8 pp mAP50 improvement.

Junjie Fan, Yijun Mai, Linduo Wei et al.

An SDT extension combining geometric vision, class-wise fusion, and affective priors reaches 75.93% and 74.11% weighted F1 on MELD and IEMOCAP.

Oriol Marín, Roger Marí, Gloria Haro et al.

TeethGNN fuses CBCT, ANB/MP-FH morphology, and graph calibration, reaching 77.08% accuracy and 89.61% AUC.

Zhichun Jin, Zhicheng He, Hao Xu et al.

SocialRL combines multi-turn PPO with dynamic process rewards, raising goal achievement by an average of 9.2 percentage points.

Jianing Wang, Xintao Wang, Aili Chen et al.

MotionBlind reveals Video-LLMs' inability to accurately understand motion in videos; only Gemini3.1 Pro passes.

Dhairya Bhatia, Bishoy Galoaa, Oliver Fritsche et al.

VANTAGE-Bench evaluates the Infrastructure AI gap in VLMs, revealing deficiencies in event verification and temporal localization tasks.

Zaid Pervaiz Bhat, Nimra Nayyar, Arihant Jain et al.

ActionSplice uses Counterfactual State Transport to enable in-flight action editing, reducing LPIPS by up to 75.9% and speeding up sampling 2.73×.

Pardis Taghavi, Tingyu Guo, Jonas Lossner et al.