Multiphase-Diff uses diffusion models to generate high-contrast multiphase physical system samples, significantly improving physical and distributional fidelity.

Yining Huang, Zhenyu Liang

Proposes a depth-constrained adaptation framework, analyzing encoder-decoder regions' impact on forgetting, with key focus on shallow encoder and deep decoder layers.

Amal Saqib, Tausifa Jan Saleem, Numan Saeed et al.

AutoDesign employs a meta-harness optimization framework to recursively improve long-horizon multimodal design quality, achieving a top score of 78.32.

Yaxin Luo, Haobin Jiang, Jialv Zou et al.

OmniScientist employs multimodal perception and a multi-agent architecture to automate multidisciplinary research from raw evidence to publication, achieving an average score of 6.3.

Bobo Li, Hao Fei, Tianjie Ju et al.

HumanScore, a preference-trained metric, uses 153 hours of diverse motion data to improve human-aligned humanoid motion tracking evaluation.

Dairu Liu, Zekun Qi, Jiayu Zeng et al.

PlayWorld employs multi-modal agent players to evaluate 171 scenarios, revealing current world models' weaknesses in spatial consistency and persistent state evolution.

Kaixin Ding, Xi Chen, Minghong Cai et al.

Vero benchmark assesses AI's ability to jointly generate verified multi-module repositories; strongest model solves 27 out of 43 instances.

Zhe Ye, Hantao Lou, Yuechun Sun et al.

Intervention-aware clinical world model encodes patient anatomy into a 3D latent state, evolving it through asynchronous post-intervention events for long-term recurrence prediction, achieving AUROC 0.756 on DECAAF-II.

Yunsung Chung, Yingshuo Liu, Abboud F. Hassan et al.

TabSOM leverages Self-Organizing Maps to encode tabular data into images, achieving top classification performance and interpretability across multiple datasets.

David Chushig-Muzo, María Ángeles Rodríguez de Cara, Eva Milara et al.

Using the SbCD model, complete crystal structures are generated via Markovian jump diffusion, excelling on MP20.

Van Khoa Nguyen, Alexandros Kalousis

The EDD framework employs explicit environmental-social disentanglement for continual learning of socially appropriate robot actions, outperforming state-of-the-art baselines.

Rafal Robert Karpinski, Fethiye Irmak Dogan, Nikhil Churamani et al.

ContactGuard employs action-conditioned latent world models to predict imminent contact failures, enabling pre-contact aborts and enhancing robotic manipulation safety.

Gehan Zheng, Matthew Johnson-Roberson, Weiming Zhi

Seeker learns control-relevant visual regions from action supervision, boosting robot task success from 48.3% to 76.7%.

Zheyu Zhuang, Ruiyu Wang, Nick Heppert et al.

Proposed a framework to evaluate long-horizon AI research, analyzing 7 models across 36 tasks.

Yiwei Li, Wanli Yang, Hexiang Tan et al.

Proposes actuation-perception co-design with micro-deformation sensing and short-thick-beam modeling, achieving real-time shape and force perception in compact capstan-driven continuum robots.

Gang Zhang, Yufu Qiu, Junyan Yan et al.

RippleMem uses event-centric memory graphs and associative recall, improving accuracy by 11.87% and reducing graph construction costs by 30x.

Jingbo Ji, Lingyi Li, Xilong Cheng et al.

CogChat uses Heterogeneous Graph Transformer and knowledge graphs to enhance conversational AI for cognitive grounding in design generation.

Jiin Choi, Kyung Hoon Hyun

Introduces Sinkhorn linearization and spectral proxy, unifying statistical and algorithmic theory of feature-parameterized inverse OT with a core spectral bound.

Han Dong, Jiaming Li, Yongqiang Gong et al.

Combining ridge regression with minimum description length enables low false positive neighborhood selection in high-dimensional networks.

Lourens Waldorp

Uniform Herding dynamically refreshes exemplars in feature space, achieving 44.00% accuracy on CIFAR-100, outperforming iCaRL.

Krishna Subedi