Geometrical properties of Nu support vector machines with different norms

Kazushi Ikeda, Noboru Murata

    Research output: Contribution to journalArticle

    17 Citations (Scopus)

    Abstract

    By employing the L1 or L norms in maximizing margins, support vector machines (SVMs) result in a linear programming problem that requires a lower computational load compared to SVMs with the L2 norm. However, how the change of norm affects the generalization ability of SVMs has not been clarified so far except for numerical experiments. In this letter, the geometrical meaning of SVMs with the Lp norm is investigated, and the SVM solutions are shown to have rather little dependency on p.

    Original languageEnglish
    Pages (from-to)2508-2529
    Number of pages22
    JournalNeural Computation
    Volume17
    Issue number11
    DOIs
    Publication statusPublished - 2005 Nov

    Fingerprint

    Support vector machines
    Linear Programming
    Linear programming
    Support Vector Machine
    Experiments

    ASJC Scopus subject areas

    • Control and Systems Engineering
    • Artificial Intelligence
    • Neuroscience(all)

    Cite this

    Geometrical properties of Nu support vector machines with different norms. / Ikeda, Kazushi; Murata, Noboru.

    In: Neural Computation, Vol. 17, No. 11, 11.2005, p. 2508-2529.

    Research output: Contribution to journalArticle

    @article{1f710f07c4e746aaa45eca9cacf6e7f7,
    title = "Geometrical properties of Nu support vector machines with different norms",
    abstract = "By employing the L1 or L∞ norms in maximizing margins, support vector machines (SVMs) result in a linear programming problem that requires a lower computational load compared to SVMs with the L2 norm. However, how the change of norm affects the generalization ability of SVMs has not been clarified so far except for numerical experiments. In this letter, the geometrical meaning of SVMs with the Lp norm is investigated, and the SVM solutions are shown to have rather little dependency on p.",
    author = "Kazushi Ikeda and Noboru Murata",
    year = "2005",
    month = "11",
    doi = "10.1162/0899766054796897",
    language = "English",
    volume = "17",
    pages = "2508--2529",
    journal = "Neural Computation",
    issn = "0899-7667",
    publisher = "MIT Press Journals",
    number = "11",

    }

    TY - JOUR

    T1 - Geometrical properties of Nu support vector machines with different norms

    AU - Ikeda, Kazushi

    AU - Murata, Noboru

    PY - 2005/11

    Y1 - 2005/11

    N2 - By employing the L1 or L∞ norms in maximizing margins, support vector machines (SVMs) result in a linear programming problem that requires a lower computational load compared to SVMs with the L2 norm. However, how the change of norm affects the generalization ability of SVMs has not been clarified so far except for numerical experiments. In this letter, the geometrical meaning of SVMs with the Lp norm is investigated, and the SVM solutions are shown to have rather little dependency on p.

    AB - By employing the L1 or L∞ norms in maximizing margins, support vector machines (SVMs) result in a linear programming problem that requires a lower computational load compared to SVMs with the L2 norm. However, how the change of norm affects the generalization ability of SVMs has not been clarified so far except for numerical experiments. In this letter, the geometrical meaning of SVMs with the Lp norm is investigated, and the SVM solutions are shown to have rather little dependency on p.

    UR - http://www.scopus.com/inward/record.url?scp=25144470140&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=25144470140&partnerID=8YFLogxK

    U2 - 10.1162/0899766054796897

    DO - 10.1162/0899766054796897

    M3 - Article

    VL - 17

    SP - 2508

    EP - 2529

    JO - Neural Computation

    JF - Neural Computation

    SN - 0899-7667

    IS - 11

    ER -