Svm Classifiers with Moderated Outputs for Automatic Classification in Molecular Biology

Madevska-Bogdanova, Ana and Nikolikj, Dragan (2002) Svm Classifiers with Moderated Outputs for Automatic Classification in Molecular Biology. In: Proceedings of the Third Conference on Informatics and Information Technology. Institute of Informatics, Faculty of Natural Sciences and Mathematics, Ss. Cyril and Methodius University in Skopje, Macedonia, Skopje, Macedonia, pp. 90-96. ISBN 9989-668-36-1

[img]
Preview
Text
9989-668-36-1_pp90-96.pdf

Download (441kB) | Preview
Official URL: http://ciit.finki.ukim.mk

Abstract

We present an alternative way of interpreting and modifying the outputs of the Support Vector Machine (SVM) classifiers – method MSVMO (Modified SVM Outputs). Stemming from the geometrical interpretation of the SVM outputs as a distance of individual patterns from the hyperplane, allows us to calculate its posterior probability i.e. to construct a probability-based measure of belonging to one of the classes, depending on the vector's relative distance from the hyperplane. We illustrate the results by providing suitable analysis of three classification problems and comparing them with an already published method for modifying SVM outputs.

Item Type: Book Section
Uncontrolled Keywords: Support Vector Machines; pattern classification; modified outputs; post-processing; posterior probability
Subjects: International Conference on Informatics and Information Technologies > Artificial Intelligence
International Conference on Informatics and Information Technologies > Intelligent Systems
Depositing User: Vangel Ajanovski
Date Deposited: 28 Oct 2016 00:15
Last Modified: 28 Oct 2016 00:15
URI: http://eprints.finki.ukim.mk/id/eprint/11031

Actions (login required)

View Item View Item