a new paper on SVM

[Chih-Jen Lin]

Dear Colleagues:

We announce a new paper on support vector machines:
A comparison on methods for multi-class support vector machines
by
Chih-Wei Hsu and Chih-Jen Lin.
http://www.csie.ntu.edu.tw/~cjlin/papers/multisvm.ps.gz

Abstract:
Support vector machines (SVM) was originally designed for binary classification.
How to effectively extend it for multi-class classification is still
an on-going research issue. Several methods have been proposed where
typically we construct a multi-class classifier by combining several 
binary classifiers. Some authors also proposed methods that consider all 
classes of data at once. As it is computationally more expensive on 
solving multi-class problems, comparisons on these methods using large-scale
problems have not been seriously conducted. Especially for methods 
solving multi-class SVM in one step, a much larger optimization
problem  is required so  up to now experiments are limited to small data sets. 
In this paper we give  decomposition implementation for 
two such  ``all-together" methods: (Vapnik 98; Weston and Watkins 1998) and 
(Crammer and Singer 2000). We then compare their performance with three methods
based on binary classification:  ``one-against-all,'' ``one-against-one,''
and DAGSVM (Platt et al. 2000). Our experiments indicate that the ``one-against-one'' 
and DAG methods are more suitable for practical use than the
other methods. Results also show that for large problems
the  method by considering all data at once  in general needs
fewer support vectors.


Any comments are very welcome.

Best
Chih-Jen Lin
Dept. of Computer Science
National Taiwan Univ.