New Methods for Learning to Cluster

Professor Fei Sha
Professor, University of Southern California
Given on: Mar. 20th, 2014


Clustering is an important component for exploratory data analysis. In this talk, I will describe two new ideas for unsupervised clustering and learning distance metrics. The first idea is pertinent to information-theoretical clustering, which surprisingly gives rise to an interesting negative result. The second idea leads to a method for learning non-metric distances, expanding our ability in modeling (dis)similarity relationships that do not observe the triangle inequality.

This talk is based on joint work with my collaborators Aram Galstyan (USCISI), Greg Ver Steeg (USCISI), Simon DeDeo (Santa Fe Inst.), and my students (Soravit Changpinyo, and Kuan Liu)


Fei Sha is an associate professor at the University of Southern California, Dept. of Computer Science. His primary research interests are machine learning and application to speech and language processing, computer vision, and robotics. He had won outstanding student paper awards at NIPS 2006 and ICML 2004. He was selected as a Sloan Research Fellow in 2013, won an Army Research Office Young Investigator Award in 2012, and was a member of DARPA 2010 Computer Science Study Panel. He has a Ph.D (2007) from Computer and Information Science from U. of Pennsylvania and B.Sc and M.Sc from Southeast University (Nanjing, China)