Benchmarking the n-tuple classifier with statlog dataset

Richard Rohwer, Michal Morciniec

    Research output: Chapter in Book/Report/Conference proceeding › Chapter

    Abstract

    The n-tuple recognition method was tested on 11 large real-world data sets and its performance compared to 23 other classification algorithms. On 7 of these, the results show no systematic performance gap between the n-tuple method and the others. Evidence was found to support a possible explanation for why the n-tuple method yields poor results for certain datasets. Preliminary empirical results of a study of the confidence interval (the difference between the two highest scores) are also reported. These suggest a counter-intuitive correlation between the confidence interval distribution and the overall classification performance of the system.
    Original languageEnglish
    Title of host publicationProceedings of the Second Weightless Neural Network Workshop 1995, Computing with Logical Neurons
    EditorsD. L. Bisset
    Place of PublicationCanterbury
    PublisherUniversity of Kent
    Pages29-34
    Number of pages6
    Publication statusPublished - 1995

    Fingerprint

    benchmarking
    confidence interval
    method

    Keywords

    • n-tuple recognition
    • algorithms
    • datasets

    Cite this

    Rohwer, R., & Morciniec, M. (1995). Benchmarking the n-tuple classifier with statlog dataset. In D. L. Bisset (Ed.), Proceedings of the Second Weightless Neural Network Workshop 1995, Computing with Logical Neurons (pp. 29-34). Canterbury: University of Kent.
    Rohwer, Richard ; Morciniec, Michal. / Benchmarking the n-tuple classifier with statlog dataset. Proceedings of the Second Weightless Neural Network Workshop 1995, Computing with Logical Neurons. editor / D. L. Bisset. Canterbury : University of Kent, 1995. pp. 29-34
    @inbook{c14ab04117a846299a5423d3ef6ae33d,
    title = "Benchmarking the n-tuple classifier with statlog dataset",
    abstract = "The n-tuple recognition method was tested on 11 large real-world data sets and its performance compared to 23 other classification algorithms. On 7 of these, the results show no systematic performance gap between the n-tuple method and the others. Evidence was found to support a possible explanation for why the n-tuple method yields poor results for certain datasets. Preliminary empirical results of a study of the confidence interval (the difference between the two highest scores) are also reported. These suggest a counter-intuitive correlation between the confidence interval distribution and the overall classification performance of the system.",
    keywords = "n-tuple recognition, algorithms, datasets",
    author = "Richard Rohwer and Michal Morciniec",
    year = "1995",
    language = "English",
    pages = "29--34",
    editor = "Bisset, {D. L.}",
    booktitle = "Proceedings of the Second Weightless Neural Network Workshop 1995, Computing with Logical Neurons",
    publisher = "University of Kent",

    }

    Rohwer, R & Morciniec, M 1995, Benchmarking the n-tuple classifier with statlog dataset. in DL Bisset (ed.), Proceedings of the Second Weightless Neural Network Workshop 1995, Computing with Logical Neurons. University of Kent, Canterbury, pp. 29-34.

    Benchmarking the n-tuple classifier with statlog dataset. / Rohwer, Richard; Morciniec, Michal.

    Proceedings of the Second Weightless Neural Network Workshop 1995, Computing with Logical Neurons. ed. / D. L. Bisset. Canterbury : University of Kent, 1995. p. 29-34.

    Research output: Chapter in Book/Report/Conference proceeding › Chapter

    TY - CHAP

    T1 - Benchmarking the n-tuple classifier with statlog dataset

    AU - Rohwer, Richard

    AU - Morciniec, Michal

    PY - 1995

    Y1 - 1995

    N2 - The n-tuple recognition method was tested on 11 large real-world data sets and its performance compared to 23 other classification algorithms. On 7 of these, the results show no systematic performance gap between the n-tuple method and the others. Evidence was found to support a possible explanation for why the n-tuple method yields poor results for certain datasets. Preliminary empirical results of a study of the confidence interval (the difference between the two highest scores) are also reported. These suggest a counter-intuitive correlation between the confidence interval distribution and the overall classification performance of the system.

    AB - The n-tuple recognition method was tested on 11 large real-world data sets and its performance compared to 23 other classification algorithms. On 7 of these, the results show no systematic performance gap between the n-tuple method and the others. Evidence was found to support a possible explanation for why the n-tuple method yields poor results for certain datasets. Preliminary empirical results of a study of the confidence interval (the difference between the two highest scores) are also reported. These suggest a counter-intuitive correlation between the confidence interval distribution and the overall classification performance of the system.

    KW - n-tuple recognition

    KW - algorithms

    KW - datasets

    M3 - Chapter

    SP - 29

    EP - 34

    BT - Proceedings of the Second Weightless Neural Network Workshop 1995, Computing with Logical Neurons

    A2 - Bisset, D. L.

    PB - University of Kent

    CY - Canterbury

    ER -

    Rohwer R, Morciniec M. Benchmarking the n-tuple classifier with statlog dataset. In Bisset DL, editor, Proceedings of the Second Weightless Neural Network Workshop 1995, Computing with Logical Neurons. Canterbury: University of Kent. 1995. p. 29-34

    Access to Document