Classification of volatile organic compounds with incremental SVMs and RBF networks

Zeki Erdem, Robi Polikar, Nejat Yumuşak, Fikret Gürgen

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Scopus citations

Abstract

Support Vector Machines (SVMs) have been applied to solve the classification of volatile organic compounds (VOC) data in some recent studies. SVMs provide good generalization performance in detection and classification of VOC data. However, in many applications involving VOC data, it is not unusual for additional data, which may include new classes, to become available over time, which then requires an SVM classifier that is capable of incremental learning that does not suffer from loss of previously acquired knowledge. In our previous work, we have proposed the incremental SVM approach based on Learn++.MT. In this contribution, the ability of SVMLearn ++.MT to incrementally classify VOC data is evaluated and compared against a similarly constructed Learn++.MT algorithm that uses radial basis function neural network as base classifiers.

Original languageEnglish (US)
Title of host publicationComputer and Information Sciences - ISCIS 2005 - 20th International Symposium, Proceedings
Pages322-331
Number of pages10
DOIs
StatePublished - Dec 1 2005
Event20th International Symposium on Computer and Information Sciences, ISCIS 2005 - Istanbul, Turkey
Duration: Oct 26 2005Oct 28 2005

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume3733 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other20th International Symposium on Computer and Information Sciences, ISCIS 2005
CountryTurkey
CityIstanbul
Period10/26/0510/28/05

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint Dive into the research topics of 'Classification of volatile organic compounds with incremental SVMs and RBF networks'. Together they form a unique fingerprint.

Cite this