Good-turing estimation for the frequentist n-tuple classifier

Michal Morciniec, Richard Rohwer

    Research output: Chapter in Book/Report/Conference proceedingChapter

    Abstract

    We present results concerning the application of the Good-Turing (GT) estimation method to the frequentist n-tuple system. We show that the Good-Turing method can, to a certain extent rectify the Zero Frequency Problem by providing, within a formal framework, improved estimates of small tallies. We also show that it leads to better tuple system performance than Maximum Likelihood estimation (MLE). However, preliminary experimental results suggest that replacing zero tallies with an arbitrary constant close to zero before MLE yields better performance than that of GT system.
    Original languageEnglish
    Title of host publicationWeightless Neural Network Workshop'95, Computing with Logical Neurons
    EditorsDavid Bisset
    Place of PublicationCanterbury
    PublisherUniversity of Kent
    Pages93-102
    Number of pages10
    Publication statusPublished - Sep 1995
    EventProceedings of the Weightless Neural Network Workshop 1995, Computing with Logical Neurons -
    Duration: 1 Sep 19951 Sep 1995

    Workshop

    WorkshopProceedings of the Weightless Neural Network Workshop 1995, Computing with Logical Neurons
    Period1/09/951/09/95

    Fingerprint

    Maximum likelihood estimation
    Classifiers

    Keywords

    • good-turing
    • zero frequency
    • estimates
    • maximum likelihood estimation

    Cite this

    Morciniec, M., & Rohwer, R. (1995). Good-turing estimation for the frequentist n-tuple classifier. In D. Bisset (Ed.), Weightless Neural Network Workshop'95, Computing with Logical Neurons (pp. 93-102). Canterbury: University of Kent.
    Morciniec, Michal ; Rohwer, Richard. / Good-turing estimation for the frequentist n-tuple classifier. Weightless Neural Network Workshop'95, Computing with Logical Neurons. editor / David Bisset. Canterbury : University of Kent, 1995. pp. 93-102
    @inbook{2de7fab561b64d3cbffa55c5884657dd,
    title = "Good-turing estimation for the frequentist n-tuple classifier",
    abstract = "We present results concerning the application of the Good-Turing (GT) estimation method to the frequentist n-tuple system. We show that the Good-Turing method can, to a certain extent rectify the Zero Frequency Problem by providing, within a formal framework, improved estimates of small tallies. We also show that it leads to better tuple system performance than Maximum Likelihood estimation (MLE). However, preliminary experimental results suggest that replacing zero tallies with an arbitrary constant close to zero before MLE yields better performance than that of GT system.",
    keywords = "good-turing, zero frequency, estimates, maximum likelihood estimation",
    author = "Michal Morciniec and Richard Rohwer",
    year = "1995",
    month = "9",
    language = "English",
    pages = "93--102",
    editor = "David Bisset",
    booktitle = "Weightless Neural Network Workshop'95, Computing with Logical Neurons",
    publisher = "University of Kent",

    }

    Morciniec, M & Rohwer, R 1995, Good-turing estimation for the frequentist n-tuple classifier. in D Bisset (ed.), Weightless Neural Network Workshop'95, Computing with Logical Neurons. University of Kent, Canterbury, pp. 93-102, Proceedings of the Weightless Neural Network Workshop 1995, Computing with Logical Neurons, 1/09/95.

    Good-turing estimation for the frequentist n-tuple classifier. / Morciniec, Michal; Rohwer, Richard.

    Weightless Neural Network Workshop'95, Computing with Logical Neurons. ed. / David Bisset. Canterbury : University of Kent, 1995. p. 93-102.

    Research output: Chapter in Book/Report/Conference proceedingChapter

    TY - CHAP

    T1 - Good-turing estimation for the frequentist n-tuple classifier

    AU - Morciniec, Michal

    AU - Rohwer, Richard

    PY - 1995/9

    Y1 - 1995/9

    N2 - We present results concerning the application of the Good-Turing (GT) estimation method to the frequentist n-tuple system. We show that the Good-Turing method can, to a certain extent rectify the Zero Frequency Problem by providing, within a formal framework, improved estimates of small tallies. We also show that it leads to better tuple system performance than Maximum Likelihood estimation (MLE). However, preliminary experimental results suggest that replacing zero tallies with an arbitrary constant close to zero before MLE yields better performance than that of GT system.

    AB - We present results concerning the application of the Good-Turing (GT) estimation method to the frequentist n-tuple system. We show that the Good-Turing method can, to a certain extent rectify the Zero Frequency Problem by providing, within a formal framework, improved estimates of small tallies. We also show that it leads to better tuple system performance than Maximum Likelihood estimation (MLE). However, preliminary experimental results suggest that replacing zero tallies with an arbitrary constant close to zero before MLE yields better performance than that of GT system.

    KW - good-turing

    KW - zero frequency

    KW - estimates

    KW - maximum likelihood estimation

    M3 - Chapter

    SP - 93

    EP - 102

    BT - Weightless Neural Network Workshop'95, Computing with Logical Neurons

    A2 - Bisset, David

    PB - University of Kent

    CY - Canterbury

    ER -

    Morciniec M, Rohwer R. Good-turing estimation for the frequentist n-tuple classifier. In Bisset D, editor, Weightless Neural Network Workshop'95, Computing with Logical Neurons. Canterbury: University of Kent. 1995. p. 93-102