DrivingScene employs a static-to-dynamic two-stage training with residual scene flow, achieving real-time high-fidelity 4D scene reconstruction from two images.

Qirui Hou, Wenzhang Sun, Chang Zeng et al.

SpikePool uses pooling attention for event-driven spiking transformers, reducing training time by 42.5% and inference time by 32.8%.

Donghyun Lee, Alex Sima, Yuhang Li et al.

OneLife framework learns symbolic world models in Crafter-OO via unguided exploration, outperforming baselines.

Zaid Khan, Archiki Prasad, Elias Stengel-Eskin et al.

Using singular learning theory to extend the minimum description length principle, this study finds local learning coefficient closely correlates with neural network compressibility.

Einar Urdshals, Edmund Lau, Jesse Hoogland et al.

GAR improves theorem proving efficiency via adversarial RL, achieving a 4.20% gain on MiniF2F-Test.

Ruida Wang, Jiarui Yao, Rui Pan et al.

Proposed OneRec-Think integrates dialogue, reasoning, and personalized recommendation, using multimodal alignment and reinforcement learning to enhance interpretability and industrial deployment.

Zhanyu Liu, Shiyao Wang, Xingmei Wang et al.

EvoCAD combines vision language models and evolutionary optimization to generate CAD code, significantly improving topological correctness.

Tobias Preintner, Weixuan Yuan, Adrian König et al.

DemoHLM achieves generalizable humanoid loco-manipulation from a single demonstration, validated across ten tasks with robust performance.

Yuhui Fu, Feiyang Xie, Chaoyi Xu et al.

ProofFlow constructs dependency graphs for faithful proof autoformalization, achieving PROOFSCORE 0.545, surpassing baselines.

Rafael Cabral, Tuan Manh Do, Xuejun Yu et al.

HatLLM uses hierarchical attention masks to strengthen cross-item collaboration, achieving a 9.13% average gain.

Yu Cui, Feng Liu, Jiawei Chen et al.

PaperArena creates a challenging benchmark for tool-augmented cross-paper reasoning, with an average accuracy of 38.78%, dropping to 18.47% on hard subsets.

Daoyu Wang, Mingyue Cheng, Shuo Yu et al.

Proposes an information-theoretic framework using predictive information and learning curves to assess the inherent learnability of sequential data.

Mario Morawski, Anais Despres, Rémi Rehm

Proposes a novel ML framework for detecting corner cases in autonomous driving, enhancing detection accuracy.

Sebastian Schmidt, Julius Körner, Stephan Günnemann

X-VLA employs soft-prompted Transformer, integrating diverse robotic datasets for scalable cross-embodiment learning with minimal parameter tuning.

Jinliang Zheng, Jianxiong Li, Zhihao Wang et al.

Diffusion-inspired masked fine-tuning enhances knowledge injection in autoregressive LLMs without paraphrases.

Xu Pan, Ely Hahami, Jingxuan Fan et al.

DELTA employs a layer-aware token selection mechanism, reducing attended tokens by up to 4.25× and speeding up inference by 1.54× while maintaining accuracy.

Hossein Entezari Zarch, Lei Gao, Chaoyi Jiang et al.

Proposes SIMBA UQ, a similarity-based framework for black-box uncertainty quantification, improving calibration on QA, summarization, and SQL tasks.

Debarun Bhattacharjya, Balaji Ganesan, Junkyu Lee et al.

FIPER combines RND and action-chunk entropy for early failure prediction of generative robot policies without failure data, outperforming existing methods.

Ralf Römer, Adrian Kobras, Luca Worbis et al.

Proposes a domain-conditioned, image-free instance generation method that boosts ILR performance by 15% on average across seven benchmarks.

Yankun Wu, Zakaria Laskar, Giorgos Kordopatis-Zilos et al.

Proposes Hierarchical Semantic RL (HSRL) using fixed semantic space to handle large, dynamic action spaces in recommendation systems.

Minmao Wang, Xingchen Liu, Shijie Yi et al.