UCL logo

UCL Discovery

UCL home » Library Services » Electronic resources » UCL Discovery

From margins to probabilities in multiclass learning problems

Passerini, A; Pontil, M; Frasconi, P; (2002) From margins to probabilities in multiclass learning problems. In: VanHarmelen, F, (ed.) ECAI 2002: 15TH EUROPEAN CONFERENCE ON ARTIFICIAL INTELLIGENCE, PROCEEDINGS. (pp. 400 - 404). I O S PRESS

Full text not available from this repository.

Abstract

We study the problem of multiclass classification within the framework of error correcting output codes (ECOC) using margin-based binary classifiers. An important open problem in this context is how to measure the distance between class codewords and the outputs of the classifiers. In this paper we propose a new decoding function that combines the margins through an estimate of their class conditional probabilities. We report experiments using support vector machines as the base binary classifiers, showing the advantage of the proposed decoding function over other functions of the margin commonly used in practice. We also present new theoretical results bounding the leave-one-out error of ECOC of kernel machines, which can be used to tune kernel parameters. An empirical validation indicates that the bound leads to good estimates of kernel parameters and the corresponding classifiers attain high accuracy.

Type: Proceedings paper
Title: From margins to probabilities in multiclass learning problems
Event: 15th European Conference on Artificial Intelligence
Location: CLAUDE BERNARD UNIV, LYON, FRANCE
Dates: 2002-07-21 - 2002-07-26
ISBN: 1-58603-257-7
Keywords: machine learning, error correcting output codes, support vector machines, statistical learning theory
UCL classification: UCL > School of BEAMS > Faculty of Engineering Science > Computer Science
URI: http://discovery.ucl.ac.uk/id/eprint/163507
Downloads since deposit
0Downloads
Download activity - last month
Download activity - last 12 months
Downloads by country - last 12 months

Archive Staff Only

View Item View Item