Bo XU

Full Professor, Institute of automation, Chinese academy of science, Chinese Academy of Sciences

3 papers at NeurIPS 2025

OpenReview· Semantic Scholar· Google Scholar

Poster Session 2

2 papers

Wednesday, December 3, 2025 · 4:30 PM → 7:30 PM

Exhibit Hall C,D,E

STAIR: Addressing Stage Misalignment through Temporal-Aligned Preference Reinforcement Learning

#409 · Yao Luan, Ni Mu, Yiqin Yang, Bo XU, Qing-Shan Jia

Self-Verifying Reflection Helps Transformers with CoT Reasoning

#4005 · Zhongwei Yu, Wannian Xia, Xue Yan, Bo XU, Haifeng Zhang, Yali Du, Jun Wang

Transformers can learn self-verifying reflection without language, and reinforcement learning enhances performance through shallow statistical patterns.

Poster Session 6

1 paper

Friday, December 5, 2025 · 4:30 PM → 7:30 PM

Exhibit Hall C,D,E

DAIL: Beyond Task Ambiguity for Language-Conditioned Reinforcement Learning

#211 · Runpeng Xie, Quanwei Wang, Hao Hu, Zherui Zhou, Ni Mu, Xiyun Li, Yiqin Yang, Shuang Xu, Qianchuan Zhao, Bo XU

This work presents DAIL, a semantic aligned reinforcement learning method that improves task discrimination and generalization in language-conditioned problems.