Published multiple papers in top international conferences such as ICML, ICLR, EMNLP, and ICCV. Some papers have received oral presentations and outstanding paper awards. Research areas include neural networks learning world models, reasoning processes in large language models, multimodal scene understanding, feature contamination, SNN conversion methods, view bias in prompt learning, counterfactual inference in reinforcement learning, and hierarchical reinforcement learning.
Research Experience
Currently a postdoc at the Department of Automation, Tsinghua University, collaborating with Prof. Feng Chen. Also leads the technical team of Qianjue Robots.
Education
Received Ph.D. and B.E. degrees from the Department of Automation, Tsinghua University in 2019. Interned as an undergraduate at IVG, Tsinghua University, advised by Prof. Jiwen Lu.
Background
Research interests include representation learning and generalization, believing that a good representation is key to achieving human-level generalization. Recent research focuses on the theoretical foundations of learning world representations and out-of-distribution generalization, and translating these principles into efficient algorithms for real-world problems. Also works on reinforcement learning and embodied intelligence.