Jiu-Cheng Xie 谢九成

jiuchengxie@gmail.com; jiuchengxie@njupt.edu.cn

I am currently affiliated with the School of Automation at Nanjing University of Posts and Telecommunications. I earned my Ph.D. from the University of Macau under the supervision of Prof. Chi-Man Pun. As part of my joint Ph.D. training, I spent one year at The Hong Kong Polytechnic University under the supervision of Prof. Kin-Man Lam. Since stepping into the PI role, I have been deeply fortunate to receive invaluable guidance and mentorship from Prof. Hao Gao and Prof. Feng Xu.

My early research began with human facial attribute analysis, which taught me that robust representation learning is the core of machine perception. However, I soon hit an implicit bottleneck: purely discriminative approaches limit representation quality. Understanding and generation are two sides of the same coin. Just as a creator understands a painting far better than a critic, a model’s ability to generate data implies a deeper understanding of it. To overcome this bottleneck, I pivoted to exploring representation learning through a generative lens, aligning with Richard Feynman’s principle: “What I cannot create, I do not understand.”

Building on this generative perspective, my current applied research focuses on human-centric dynamic world modeling and simulation. This framework centers on humans and their surrounding environments, driven by three fundamental processes: reconstruction, inference, and interaction. I hope that advancing these technical frontiers will empower us with greater freedom, both materially and spiritually.

Prospective master's and undergraduate students: If you are interested in joining my research group, please email me with your CV.