Selecting training points for one-class support vector machines

Yuhua Li

    Research output: Contribution to journal › Article

    47 Citations (Scopus)

    Abstract

    This paper proposes a training points selection method for one-class support vector machines. It exploits the feature of a trained one-class SVM, which uses points only residing on the exterior region of data distribution as support vectors. Thus, the proposed training set reduction method selects the so-called extreme points which sit on the boundary of data distribution, through local geometry and k-nearest neighbours. Experimental results demonstrate that the proposed method can reduce training set considerably, while the obtained model maintains generalization capability to the level of a model trained on the full training set, but uses less support vectors and exhibits faster training speed.
    LanguageEnglish
    Pages1517-1522
    JournalPattern Recognition Letters
    Volume32
    Issue number11
    DOIs
    Publication statusPublished - 1 Aug 2011

    Fingerprint

    Support vector machines
    Geometry

    Cite this

    Li, Yuhua. / Selecting training points for one-class support vector machines. In: Pattern Recognition Letters. 2011 ; Vol. 32, No. 11. pp. 1517-1522.
    @article{58b03a1bba8c460590872ca2e3ac69b8,
    title = "Selecting training points for one-class support vector machines",
    abstract = "This paper proposes a training points selection method for one-class support vector machines. It exploits the feature of a trained one-class SVM, which uses points only residing on the exterior region of data distribution as support vectors. Thus, the proposed training set reduction method selects the so-called extreme points which sit on the boundary of data distribution, through local geometry and k-nearest neighbours. Experimental results demonstrate that the proposed method can reduce training set considerably, while the obtained model maintains generalization capability to the level of a model trained on the full training set, but uses less support vectors and exhibits faster training speed.",
    author = "Yuhua Li",
    year = "2011",
    month = "8",
    day = "1",
    doi = "10.1016/j.patrec.2011.04.013",
    language = "English",
    volume = "32",
    pages = "1517--1522",
    journal = "Pattern Recognition Letters",
    issn = "0167-8655",
    publisher = "Elsevier",
    number = "11",

    }

    Li, Y 2011, 'Selecting training points for one-class support vector machines', Pattern Recognition Letters, vol. 32, no. 11, pp. 1517-1522. https://doi.org/10.1016/j.patrec.2011.04.013

    Selecting training points for one-class support vector machines. / Li, Yuhua.

    In: Pattern Recognition Letters, Vol. 32, No. 11, 01.08.2011, p. 1517-1522.

    Research output: Contribution to journal › Article

    TY - JOUR

    T1 - Selecting training points for one-class support vector machines

    AU - Li, Yuhua

    PY - 2011/8/1

    Y1 - 2011/8/1

    N2 - This paper proposes a training points selection method for one-class support vector machines. It exploits the feature of a trained one-class SVM, which uses points only residing on the exterior region of data distribution as support vectors. Thus, the proposed training set reduction method selects the so-called extreme points which sit on the boundary of data distribution, through local geometry and k-nearest neighbours. Experimental results demonstrate that the proposed method can reduce training set considerably, while the obtained model maintains generalization capability to the level of a model trained on the full training set, but uses less support vectors and exhibits faster training speed.

    AB - This paper proposes a training points selection method for one-class support vector machines. It exploits the feature of a trained one-class SVM, which uses points only residing on the exterior region of data distribution as support vectors. Thus, the proposed training set reduction method selects the so-called extreme points which sit on the boundary of data distribution, through local geometry and k-nearest neighbours. Experimental results demonstrate that the proposed method can reduce training set considerably, while the obtained model maintains generalization capability to the level of a model trained on the full training set, but uses less support vectors and exhibits faster training speed.

    U2 - 10.1016/j.patrec.2011.04.013

    DO - 10.1016/j.patrec.2011.04.013

    M3 - Article

    VL - 32

    SP - 1517

    EP - 1522

    JO - Pattern Recognition Letters

    T2 - Pattern Recognition Letters

    JF - Pattern Recognition Letters

    SN - 0167-8655

    IS - 11

    ER -

    Li Y. Selecting training points for one-class support vector machines. Pattern Recognition Letters. 2011 Aug 1;32(11):1517-1522. https://doi.org/10.1016/j.patrec.2011.04.013

    Access to Document