|
Lian ZiHao I'm a Ph.D. student at the School of Future Technology, South China University of Technology (SCUT) in Guangzhou, where I am a member of the SCUT Machine Intelligence Lab (SMIL). My research interests lie in vision-language models, reinforcement learning, and continual learning. I am advised by Prof. Mingkui Tan. |
|
ResearchI'm interested in vision-language models, reinforcement learning, and continual learning. My research focuses on building intelligent systems that can understand visual and textual information jointly, adapt to new tasks over time, and transfer knowledge efficiently across domains. Some of my papers are highlighted. |
|
CKA-RL: Continual Knowledge Adaptation for Reinforcement Learning
Jinwu Hu*, Zihao Lian* (co-first author), Zhiquan Wen, Chenghao Li, Guohao Chen, Xutao Wen, Bin Xiao, Mingkui Tan NeurIPS, 2025 code / arxiv We propose CKA-RL, a continual reinforcement learning framework that efficiently accumulates and transfers task knowledge while mitigating catastrophic forgetting. |
|
NSG-VD: Physics-driven AI-generated Video Detection via Normalized Spatiotemporal Gradient
Shuhai Zhang*, Zihao Lian* (co-first author), Jiahao Yang, Daiyuan Li, Guoxuan Pang, Feng Liu, Bo Han, Shutao Li, Mingkui Tan NeurIPS, 2025 (Spotlight Presentation) code / arxiv We propose NSG-VD, a physics-driven AI-generated video detection method that captures spatiotemporal anomalies for accurate detection. |