Portrait of Shuxin Zheng

Shuxin Zheng

Senior Researcher


Shuxin Zheng (郑书新) joined Microsoft Research as a senior researcher in 2019.  Before that, he graduated from the USTC-MSRA joint PhD program, supervised by Prof. Tie-Yan Liu and Prof. Nenghai Yu. He is devoted to the research of deep learning algorithm, graph neural networks, and exploration in the field of science with machine learning. His research results have been published in top international conferences on machine learning such as ICML, NeurIPS, ICLR, CVPR, ECCV, and high impact factor scientific journals such as Environ. Sci. Technol., Atmos. Res. The representative works of Dr.Zheng include: he developed the molecular modeling algorithm Graphormer, and won the championship in several international competitions in AI molecular modeling and molecular dynamics simulation, by successively defeating DeepMind, Facebook AI Research and other teams;  jointly with the school of environment of Tsinghua University, he developed the AI numerical model which was adopted by China’s “14th Five-Year Plan” as the core technology for controlling air pollution and carbon emission in China. Dr.Zheng has long served as a reviewer for conferences and journals such as ICML, NeurIPS, ICLR, TPAMI, etc., and teach “Machine Learning Methods and Application Fundamentals” and “Advanced Machine Learning” at Tsinghua University and Microsoft AI School.


  • Graphormer is open srouced as General AI Molecular Simulation toolkit! [github]
  • Graphormer has won the Open Catalyst Challenge! [link]
  • Our work Mimicking atmospheric photochemical modeling with a deep neural network has been accepted by Atmospheric Research! [link]
  • Two papers have been accepted by NeurIPS 2021.
  • Our work Graphormer has shown that Transformer is all you need for Graph learning! [arXiv] [github] [blog]
  • We win the 1st place of KDD Cup 2021 OGB-LSC challenge (quantum chemistry track)! [kdd] [technical report]
  • Our paper How could Neural Networks understand Programs  has been accepted by ICML 2021. This work is our first attempt under the topic of Programming Language Processing (PLP). [arXiv] [github]
  • Our paper Cross-Iteration Batch Normalization has been accepted by CVPR 2021 and been integrated into YOLOv4. [arXiv] [github] [YOLOv4]

I’m looking for highly-motivated collaborators all the time. Please contact me if you’re interested.