Leyang Shen is a Ph.D. student at National University of Singapore, advised by Prof. Chua Tat-Seng. He received his Bachelor's degree from Harbin Institute of Technology, Shenzhen (HITSZ), in 2025, during which he conducted research under the supervision of Prof. Liqiang Nie and Prof. Rui Shao.
His research broadly lies in Multi-Agent Systems, Reinforcement Learning, and Multi-Modal Learning and currently focuses on Multi-Agent Reinforcement Learning.
Leyang Shen, Yang Zhang, Chun Kai Ling, Xiaoyan Zhao, Tat-Seng Chua
A criticality-aware reinforcement learning algorithm for long-horizon agentic reasoning that focuses training on high-criticality states.
Wei Li, Bing Hu, Rui Shao, Leyang Shen, Liqiang Nie
In this work, we propose “Fast & Slow Video-Language Thinker” as onLIne videO assistaNt, LION-FS, achieving real-time, proactive, temporally accurate, and contextually precise responses.
Leyang Shen, Gongwei Chen, Rui Shao, Weili Guan, Liqiang Nie
In this work, we proposed a mixture of multimodal experts (MoME) framework to mitigate task interference and obtain a generalist MLLM.
Gongwei Chen, Leyang Shen, Rui Shao, Xiang Deng, Liqiang Nie
In this work, we enhance MLLMs by integrating fine-grained spatial-aware visual knowledge and high-level semantic visual evidence, boosting capabilities and alleviating hallucinations.
Recent updates from my research and studies.
Our work MAFP has been released. We explore a paradigm of MAS that addresses stance entanglement of decision-making tasks.
Our work CARL has been released. We improve both the efficiency and performance of agentic RL by focusing on critical states.
Starting my PhD journey at NExT++ Lab @ National University of Singapore!
Our paper LION-FS has been accepted by CVPR 2025.
Our paper MoME has been accepted by NeurIPS 2024.
Our paper LION has been accepted by CVPR 2024.
