Math, Music, Sport and AI

Life is a boat on an ocean
Let's sail together

Zhizhong Li

Let's change the world!

I received Ph.D. degree from MM-Lab, IE department of CUHK at 2019. My research interests mainly focus on machine learning and computer vision. I also have work on graph CNN, reinforcement learning, ensemble learning and geometric optimization.
I received M.S. (Mathematics) from Peking University at 2015, and B.S. (Mathematics) and B.A (double major in English) degree from Zheng Zhou University at 2010.



Cong Ma*, Zhizhong Li*, Dahua Lin, Jiangshe Zhang. Parallel Multi-Environment Shaping Algorithm for Complex Multi-step Task. Neurocomputing (2020).


Hao Sun, Zhizhong Li, Xiaotong Liu, Bolei Zhou, Dahua Lin. Policy Continuation with Hindsight Inverse Dynamics; NeurIPS 2019.


Sijie Yan*, Zhizhong Li*, Yuanjun Xiong, Huahan Yan, Dahua Lin. Convolutional Sequence Generation for Skeleton-Based Action Synthesis; ICCV 2019.


Zhizhong Li. A Stroke of Genius: Generating Images. In Xiao’ou Tang and Yukun Chen, editors, Fundamentals of Artificial Intelligence (Senior High School Edition), chapter 8, pages 135–152. East China Normal University Press and The Commercial Press, 2018. [Link]


Zhizhong Li and Dahua Lin. Integrating Specialized Classifiers Based on Continuous Time Markov Chain. In Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence (IJCAI), pages 2244–2251, 2017. [paper] [code]


Xingcheng Zhang*, Zhizhong Li*, Chen Change Loy, and Dahua Lin. PolyNet: A Pursuit of Structural Diversity in Very Deep Networks. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 718–726, 2017. [paper] [repo]


Zhizhong Li, Deli Zhao, Zhouchen Lin, and Edward Y Chang. A New Retraction for Accelerating the Riemannian Three-Factor Low-Rank Matrix Completion Algorithm. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 4530–4538, 2015. [paper] [code]


Zhizhong Li, Deli Zhao, Zhouchen Lin, and Edward Y Chang. Determining Step Sizes in Geometric Optimization Algorithms. In 2015 IEEE International Symposium on Information Theory (ISIT), pages 1217–1221. IEEE, 2015. [paper] [code]



Lanxin Lei*, Zhizhong Li*, Dahua Lin. Biased Estimates of Advantages over Path Ensembles; Under review.


Xingxing Zou*, Zhizhong Li*, Ke Bai, Dahua Lin, Waikeung Wong. Explainable Clothing Aesthetic Evaluation with Gradient Penalty; Draft

Fanny Links


Lanxin Lei, Zhizhong Li, Dahua Lin. Experience Guided Exploration in Reinforcement Learning; Draft.


[Seminar Talk] A Brief Introduction to Graph Convolution. [ppt]


[Random Thought] Deep Self-Organized AutoEncoder (unfinished). [pdf]


[Seminar Talk] GAN & Optimal Transportation. [ppt]


[Random Thought] Geometric View of GAN and Visualization. [pdf] [ppt]


[Seminar Talk] Geometric Optimization Algorithms for Matrix Completion. [ppt]


李治中;杨明珊;邹澎;静电场第一类边值外问题的有限差分法求解[A];教育部中南地区高等学校电子电气基础课教学研究会第二十届学术年会会议论文集(下册)[C];2010年 [paper]


[B.A. Thesis] Discovering Cultural Gaps from Translated Works --- with the Case Study of Davis’s Translated work: Chinese Novels [pdf]

06 Feb 2019

Clean Chinese Rare Book Images with GAN

The digitalized Chinese rare books often has bad visual quality due to yellowish, translucent pages. We use CycleGAN with a novel edge-keeping loss term to remove artifacts and improve the image quality.

21 May 2018

Kong Fu