Systems Seminar

A Statistical View of Support Vector Machines

Prof. Yi Lin
UW Dept. of Statistics

Abstract

Support vector machine for pattern recognition has undergone rapid development in recent years. Much of the progress has been made in the area of developing algorthms for support vector methodology. The existing statistical theory of support vector machines does not provide a satisfactory explanation why support vector machines performance well in real world problems. In this talk, the basics of support vector machines are reviewed, and a connection between support vector machines and the Bayes rule of classification is established. This explains the good performance of support vector machines, and sheds light on how support vector machines should be used in non-standard situations.

Time and Place: Wed., Feb. 2, 3:30-4:30 pm in 4610 Engr. Hall.

SYSTEMS SEMINAR WEB PAGE: http://www.cae.wisc.edu/~gubner/seminar/