Analysis of diversity methods for evolutionary multi-objective ensemble classifiers

Stefan Oehmcke*, Justin Heinermann, Oliver Kramer

*Corresponding author af dette arbejde
    2 Citationer (Scopus)

    Abstract

    Ensemble classifiers are strong and robust methods for classification and regression tasks. Considering the balance between runtime and classifier accuracy the learning problem becomes a multi-objective optimization problem. In this work, we propose an evolutionary multiobjective algorithm based on non-dominated sorting that balances runtime and accuracy properties of nearest neighbor classifier ensembles and decision tree ensembles. We identify relevant ensemble parameters with a significant impact on the accuracy and runtime. In the experimental part of this paper, we analyze the behavior on typical classification benchmark problems.

    OriginalsprogEngelsk
    TitelApplications of Evolutionary Computation - 18th European Conference, EvoApplications 2015, Proceedings
    RedaktørerGiovanni Squillero, Antonio M. Mora
    Antal sider12
    ForlagSpringer Verlag,
    Publikationsdato1 jan. 2015
    Sider567-578
    ISBN (Elektronisk)9783319165486
    DOI
    StatusUdgivet - 1 jan. 2015
    Begivenhed18th European Conference on the Applications of Evolutionary Computation, EvoApplications 2015 - Copenhagen, Danmark
    Varighed: 8 apr. 201510 apr. 2015

    Konference

    Konference18th European Conference on the Applications of Evolutionary Computation, EvoApplications 2015
    Land/OmrådeDanmark
    ByCopenhagen
    Periode08/04/201510/04/2015
    SponsorInstitute for Informatics and Digital Innovation, National Museum of Denmark, The World Federation on Soft Computing
    NavnLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    Vol/bind9028
    ISSN0302-9743

    Fingeraftryk

    Dyk ned i forskningsemnerne om 'Analysis of diversity methods for evolutionary multi-objective ensemble classifiers'. Sammen danner de et unikt fingeraftryk.

    Citationsformater