Title: Statistical Relational Learning from Internet Data
Dr. Shenghuo Zhu
NEC Laboratories America
Date: September 25th, 2009
Location: 3105 Egr (CSE conference room)
Host: Rong Jin
Internet services generate various types of data, such as web pages with hyperlinks, customer purchase, click stream, rating, etc. Many tasks require learning from these data, for example, recommendation, ranking, customer targeting, personalization, link prediction. We explore generic statistical relational models for these data. We focus on the issues: 1) the computational complexity of learning; 2) how to efficiently incorporate auxiliary information; 3) the interpretability of models.
Bio and background:
Shenghuo Zhu received his Ph.D. in Computer Science from University of Rochester in 2003. He worked on customer behavior research in Amazon.com between 2003 and 2004. He has been working in NEC Laboratories America since 2004. Currently, he leads a team to innovate the technologies to applicable to analyze relational data of social networks, recommendation systems, etc. His main research interests are in various areas of information retrieval, machine learning and data mining.