Home | Repositories | Statistics | About



Subject: Support Vector Machines; pattern classification; modified outputs; post-processing; posterior probability


Year: 2002


Type: Proceedings



Title: Svm Classifiers with Moderated Outputs for Automatic Classification in Molecular Biology


Author: Madevska Bogdanova, Ana
Author: Nikolikj, D



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 probabilitybased 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.


Publisher: Institute of Informatics, Faculty of Natural Sciences and Mathematics, Ss. Cyril and Methodius University in Skopje, Macedonia


Relation: Third International Conference on Informatics and Information Technology



Identifier: oai:repository.ukim.mk:20.500.12188/24478
Identifier: http://hdl.handle.net/20.500.12188/24478



TitleDateViews
Svm Classifiers with Moderated Outputs for Automatic Classification in Molecular Biology200229