School of
Information Technology and Electrical Engineering

Classification learning based on information theory

Speaker: 
Baogang Hu, Professor, Senior Member - IEEEWed, 04/07/2012 - 10:00
Venue: 
78-622
Host: 
Prof Heng Tao Shen
Abstract: 

This talk introduces our recent study on information theoretical learning (ITL). By comparing with the conventional performance-based approaches, I will show that ITL presents unique features which have not been reported before in classifications. Three parts of study will be given in the talk, that is:

  • 24 information measures for objective evaluations of classifications including “error types” and “reject types”
  • theoretical comparisons between Bayesian classifiers and mutual information classifiers in cost sensitive learning and abstaining learning
  • analytical bounds between entropy and errors for Bayesian and non-Bayesian types.

Our findings confirm that ITL provides a new perspective for understanding some learning mechanisms or decision rules in our daily life. I will also present personal viewpoints on the cons and pros of ITL. 

Biography: 

Baogang Hu, Professor, Senior Member, IEEE

National Laboratory of Pattern Recognition (NLPR), Institute of Automation, Chinese Academy of Sciences, Beijing China

Baogang Hu received his Ph.D. degree in 1993 from Department of Mechanical Engineering, McMaster University, Canada. Currently, he is a professor of National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences (CAS), Beijing, China. From 2000 to 2005, he served as the Chinese Director of “Chinese-French Laboratory of Information, Automation and Applied Mathematics”(LIAMA) sponsored by CAS(China), INRIA(France), CNRS(France), and CIRAD(France). His current researches include plant growth modeling and machine learning.

Seminar Type: 

ITEE Research Seminar

Pages