Multicategory Support Vector Machines.

Date of Submission

December 2004

Date of Award

Winter 12-12-2005

Institute Name (Publisher)

Indian Statistical Institute

Document Type

Master's Dissertation

Degree Name

Master of Technology

Subject Name

Computer Science


Electronics and Communication Sciences Unit (ECSU-Kolkata)


Pal, Srimanta (ECSU-Kolkata; ISI)

Abstract (Summary of the Work)

In this dissertation first we critically review the Support Vector Machine for both two-class and multiclass problems. There have been a few attempts to deal with multiclass problems. Our analysis of a number of such methods revealed several problems associated with them. We then proposed six new methods. Three of proposed methods are modification of some existing methods, while three methods involve reformulation of Support Vector Machine. In this context we introduce a novel concept of utility of training points, which make Support Vector Machine less sensitive to outliers. We also introduce a new concept of optimal hyper-sphere to design a Support Vector Machine type of classifiers. All methods are theoretically formulated illustrated using the suitable datasets.


ProQuest Collection ID:

Control Number


Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.


This document is currently not available here.