Li, Ge


Li, Ge

Associate Professor

Research Interests: Software engineering, machine learning

Office Phone: 86-10-6275 1794-11


LI, Ge is an associate professor in the Department of Computer Science and Technology, School of EECS. He obtained his B.Sc. from Shandong University of Technology in 1999, and Ph.D. from Peking University in 2006 respectively. He had been a visiting associate professor at Artificial Intelligence Laboratory of Stanford University in 2013-2014. His current research mainly concerns applications of probabilistic methods for machine learning, including program language processing, program code generation, and natural language process.

Dr. Li is the deputy secretary general of CCF Software Engineering Society and the founder of the Software Program Generation Study Group, which includes over 100 advanced researchers in China. Dr. Li has published more than 50 research papers, and many of them are published in top-tier conferences and journals, such as AAAI, ACL, ASE, ICSE, EMNLP, CIKM, COLING and JSS. He has served as in the Technical Program Committee of many international conferences including KSEM, ICSR, UIC, ICeBE etc. He was awarded the first prize of Science and Technology Progress Award from the Department of the Ministry of Education. His undergraduate course, Introduction to Computing, was the first computer MOOC course from mainland China on Coursera, and was awarded National Excellent Course. He was also awarded Peking University Teaching Excellence Award, Beijing Educational Innovation Model, First prize of the Association of Computer Education in Beijing Colleges and Universities, First prize of Teaching Skills Competition of Beijing Universities.

His research achievements are summarized as follows:

1.  Program Language Model: Dr. Li Ge was one of the earliest researchers engaged in the study of the computer program language model based on deep neural network. He proposed a new deep convolutional neural network for the abstract syntax tree, which can extract semantic features effectively from syntax structures of program code AST. This model has been applied by many companies in multiple program analysis tasks, including code classification, malicious code analysis, code feature location, code clone detection, etc.

2.  Program Code Generation: Dr. Li Ge was also one of the earliest researchers engaged in the study of end-to-end program code generating techniques. He proposed a new program code generation framework based on an Induced Token LSTM model, which can generate follow-up program codes automatically based on the code programmers had written. This framework can be used in code completion, code bug auto fixing, etc.

3.  Natural Language Process: As applications of probabilistic methods for machine learning in NLP, Dr. Li achieved many leading research results on multiple NLP tasks, including sentiment analysis, relation extraction, efficient word embedding model, domain knowledge extraction and knowledge inference.