Song, Guojie

Associate Professor

Research Interests: Data mining

Office Phone: 86-10-6275 4785

Email: gjsong@pku.edu.cn

Song, Guojie is an associate professor in the Department of Machine Perception, School of EECS. He received the Ph.D. degree in the Department of Computer Science from Peking University, in 2004. He served as the vice Director of the Research Center of Intelligent Transportation System, at Peking University. He is currently interested in various techniques of data mining, machine learning, as well as their applications in intelligent transportation system, and social networks.

Dr. Song has published more than 60 research papers, and more than ten of them are published in top-tier conferences and journals, such as SIGKDD, AAAI, TKDE, and TPDS. He has served in the Technical Program Committee of various international conferences including ICDE, WWW, AAAI etc. He has been supported by more twenty projects, including Nature Science Foundation of China, National High Technology Research and Development Program of China and the National Science and Technology

Support Plan etc. He has been awarded the first prize of Chinese Academy of highway science and Technology Award in two years. He is the chief course instructors of national-level quality course, and achieved the first prize of the teaching achievement prizes at Peking University twice.

Dr. Song’s research achievements are summarized as follows:

1) Influence Maximization in Social Network: Influence maximization research is one of the hot topics in the field of social network analysis and data mining, which aims to extract top-k nodes which has the highest influence coverage. He proposed a community detection based and simulated annealing based influence maximization techniques to improve the efficiency more than order of magnitudes, which has been cited more than 500 times in google scholar.

2) Traffic Flow Prediction on Highway Network: Traffic flow prediction is the core the intelligent transportation system. His research focus lies in the highway networks. He proposed several new traffic prediction models which includes the DNN and Multitask learning based and local weighted learning based network wide traffic flow prediction model etc. These methods can have been used in real applications with more than two thousands economic benefit.