Attribute-Efficient Learning in Query and Mistake-Bound Models

Nader Bshouty, Lisa Hellerstein

    Research output: Contribution to journalArticle

    Abstract

    We consider the problem of attribute-efficient learning in query and mistake-bound models. Attribute-efficient algorithms make a number of queries or mistakes that is polynomial in the number of relevant variables in the target function, but only sublinear in the number of irrelevant variables. We consider a variant of the membership query model in which the learning algorithm is given as input the number of relevant variables of the target function. We show that in this model, any projection and embedding closed class of functions (including parity) that can be learned in polynomial time can be learned attribute-efficiently in polynomial time. We show that this does not hold in the randomized membership query model. In the mistake-bound model, we consider the problem of learning attribute-efficiently using hypotheses that are formulas of small depth. Our results extend the work of A. Blum, L. Hellerstein, and N. Littlestone (J. Comput. System Sci. 50 (1995), 32-40) and N. Bshouty, R. Cleve, S. Kannan, and C. Tamon (in "Proceedings, 7th Annu. ACM Workshop on Comput. Learning Theory," pp. 130-139, ACM Press, New York, 1994).

    Original languageEnglish (US)
    Pages (from-to)310-319
    Number of pages10
    JournalJournal of Computer and System Sciences
    Volume56
    Issue number3
    StatePublished - Jun 1998

    Fingerprint

    Attribute
    Query
    Polynomials
    Polynomial time
    Target
    Learning Theory
    Model
    Learning algorithms
    Parity
    Learning Algorithm
    Efficient Algorithms
    Learning
    Projection
    Closed
    Polynomial

    ASJC Scopus subject areas

    • Computational Theory and Mathematics

    Cite this

    Attribute-Efficient Learning in Query and Mistake-Bound Models. / Bshouty, Nader; Hellerstein, Lisa.

    In: Journal of Computer and System Sciences, Vol. 56, No. 3, 06.1998, p. 310-319.

    Research output: Contribution to journalArticle

    Bshouty, N & Hellerstein, L 1998, 'Attribute-Efficient Learning in Query and Mistake-Bound Models', Journal of Computer and System Sciences, vol. 56, no. 3, pp. 310-319.
    Bshouty, Nader ; Hellerstein, Lisa. / Attribute-Efficient Learning in Query and Mistake-Bound Models. In: Journal of Computer and System Sciences. 1998 ; Vol. 56, No. 3. pp. 310-319.
    @article{9975bf548cf74ebbbfcb9fe5df2a04bb,
    title = "Attribute-Efficient Learning in Query and Mistake-Bound Models",
    abstract = "We consider the problem of attribute-efficient learning in query and mistake-bound models. Attribute-efficient algorithms make a number of queries or mistakes that is polynomial in the number of relevant variables in the target function, but only sublinear in the number of irrelevant variables. We consider a variant of the membership query model in which the learning algorithm is given as input the number of relevant variables of the target function. We show that in this model, any projection and embedding closed class of functions (including parity) that can be learned in polynomial time can be learned attribute-efficiently in polynomial time. We show that this does not hold in the randomized membership query model. In the mistake-bound model, we consider the problem of learning attribute-efficiently using hypotheses that are formulas of small depth. Our results extend the work of A. Blum, L. Hellerstein, and N. Littlestone (J. Comput. System Sci. 50 (1995), 32-40) and N. Bshouty, R. Cleve, S. Kannan, and C. Tamon (in {"}Proceedings, 7th Annu. ACM Workshop on Comput. Learning Theory,{"} pp. 130-139, ACM Press, New York, 1994).",
    author = "Nader Bshouty and Lisa Hellerstein",
    year = "1998",
    month = "6",
    language = "English (US)",
    volume = "56",
    pages = "310--319",
    journal = "Journal of Computer and System Sciences",
    issn = "0022-0000",
    publisher = "Academic Press Inc.",
    number = "3",

    }

    TY - JOUR

    T1 - Attribute-Efficient Learning in Query and Mistake-Bound Models

    AU - Bshouty, Nader

    AU - Hellerstein, Lisa

    PY - 1998/6

    Y1 - 1998/6

    N2 - We consider the problem of attribute-efficient learning in query and mistake-bound models. Attribute-efficient algorithms make a number of queries or mistakes that is polynomial in the number of relevant variables in the target function, but only sublinear in the number of irrelevant variables. We consider a variant of the membership query model in which the learning algorithm is given as input the number of relevant variables of the target function. We show that in this model, any projection and embedding closed class of functions (including parity) that can be learned in polynomial time can be learned attribute-efficiently in polynomial time. We show that this does not hold in the randomized membership query model. In the mistake-bound model, we consider the problem of learning attribute-efficiently using hypotheses that are formulas of small depth. Our results extend the work of A. Blum, L. Hellerstein, and N. Littlestone (J. Comput. System Sci. 50 (1995), 32-40) and N. Bshouty, R. Cleve, S. Kannan, and C. Tamon (in "Proceedings, 7th Annu. ACM Workshop on Comput. Learning Theory," pp. 130-139, ACM Press, New York, 1994).

    AB - We consider the problem of attribute-efficient learning in query and mistake-bound models. Attribute-efficient algorithms make a number of queries or mistakes that is polynomial in the number of relevant variables in the target function, but only sublinear in the number of irrelevant variables. We consider a variant of the membership query model in which the learning algorithm is given as input the number of relevant variables of the target function. We show that in this model, any projection and embedding closed class of functions (including parity) that can be learned in polynomial time can be learned attribute-efficiently in polynomial time. We show that this does not hold in the randomized membership query model. In the mistake-bound model, we consider the problem of learning attribute-efficiently using hypotheses that are formulas of small depth. Our results extend the work of A. Blum, L. Hellerstein, and N. Littlestone (J. Comput. System Sci. 50 (1995), 32-40) and N. Bshouty, R. Cleve, S. Kannan, and C. Tamon (in "Proceedings, 7th Annu. ACM Workshop on Comput. Learning Theory," pp. 130-139, ACM Press, New York, 1994).

    UR - http://www.scopus.com/inward/record.url?scp=0032090599&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=0032090599&partnerID=8YFLogxK

    M3 - Article

    VL - 56

    SP - 310

    EP - 319

    JO - Journal of Computer and System Sciences

    JF - Journal of Computer and System Sciences

    SN - 0022-0000

    IS - 3

    ER -