Believing in the well-known dictum by Richard Feynman, "What I cannot create, I do not understand", I hold great interest in fundamental problems of deep unsupervised learning, e.g., various generation and inference methods. Based on the theoretical understanding, I’m also interested in designing novel algorithms for practical problems.
Currently, I pay more attention to the structured prediction problems on discrete data, e.g., text and graph.
[Dec. 2019]  New!!
Our paper GraphAF: a Flow-based Autoregressive Model for Molecular Graph Generation was accepted at ICLR 2020.
[Nov. 2019]  New!!
Our paper Infomax Neural Joint Source-Channel Coding via Adversarial Bit Flip was accepted at AAAI 2020.
Publication (* equal contribution)
GraphAF: a Flow-based Autoregressive Model for Molecular Graph Generation
International Conference on Learning Representations (ICLR), 2020
Infomax Neural Joint Source-Channel Coding via Adversarial Bit Flip
AAAI Conference on Artificial Intelligence (AAAI), 2020
Mong Man Wai - Hong Kong Scholarship (~6,000 USD, 15 among 6,000+ sophomore and juniorstudents in SJTU, fully funded to visit Oxford University), SJTU. 2018
Kwang-Hua Scholarship (Sophomore Ranking 3rd/104 in CSE), Kwang-Hua Foundation. 2018
Shanghai Scholarship (Ranking 2nd/104 of Freshman in CSE), Shanghai Education Ministry. 2017
Updated at Feb. 2020