KReSIT Logo Kanwal Rekhi School of Information Technology
IIT Bombay
 Current Year
 Previous Year
 Publications Archives

Home > Research > Publications 


TitleScaling multi-class Support Vector Machines using inter-class confusion
ConferenceInternational Conference on Knowledge Discovery and Data Mining KDD-2002  -   2002
Author(s)Shantanu Godbole  Sunita Sarawagi  Soumen Chakrabarti 
AbstractSupport vector machines (SVMs) excel at two-class discriminative learning problems. They often outperform genera- TIVE classiers, especially tve Bayes (NB) classier. On the other hand, generative classiers have no trouble in handling an arbitrary number of classes ehose that use inaccurate genera- tive models, such as the na.ciently, and NB classiers train much faster than SVMs owing to their extreme sim- plicity. In contrast, SVMs handle multi-class problems by learning redundant yes/no (one-vs-others) classiers for each class, further worsening the performance gap. We propose a new technique for multi-way classication which exploits the accuracy of SVMs and the speed of NB classiers. We rst use a NB classier to quickly compute a confusion ma- trix, which is used to reduce the number and complexity of the two-class SVMs that are built in the second stage. Dur- ing testing, we rst get the prediction of a NB classier and use that to selectively apply only a subset of the two-class SVMs. On standard benchmarks, our algorithm is 3 to 6 times faster than SVMs and yet matches or even exceeds their accuracy.

Printer friendly    Comment
  Copyright © 2004 KReSIT, IIT Bombay. All rights reserved sitemap    
  Kanwal Rekhi School of Information Technology, Indian Institute of Technology Bombay, Powai, Mumbai - 400 076.
+91-22-2576 7901/02. Fax: +91-22-2572 0022
Designed by Kamlesh