Zhong Li

关于

My name is Zhong Li (李 忠). I am now a researcher at Microsoft Research Asia.

I obtained the bachelor degree from Tang’aoqing Honors Program in mathematics, Jilin University, and the PhD degree from School of Mathematical Sciences, Peking University, supervised by Prof. Weinan E.

My research interests center around the mathematical theories, principles and algorithms of deep learning, including:

1. Theoretical foundations of neural networks
   • The approximation, optimization and generalization analysis of inductive biases of NNs, particularly for seq2seq/time-series modeling (e.g. RNNs, CNNs, encoder-decoders, transformers, etc)
   • Data-dependent development of machine learning
2. Mathematics and principles of AIGC
   • Theories and algorithms of diffusion models
   • Data-efficient learning, with the application to large language models

I am always seeking highly self-motivated research interns with solid mathematical foundations and skilled coding techniques. If you are interested in theories (at least, principles) of machine learning and LLMs, please feel free to contact me via the email lzhong@microsoft.com.