Estimating hidden unit number for two-layer perceptrons

Mario Gutierrez, Jennifer Wang, Robert Grondin

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    25 Citations (Scopus)

    Abstract

    The authors concentrate on estimating how many hidden units are needed for large nets. The design method is adapted from a more rigorous one developed for smaller nets where all possible input/output relationships are presented in the training vectors. It provides a 'ballpark' estimate of the minimum number of hidden units required by a two-layer perceptron for a provided training subset of the possible input/output pairs. Typically, the number of hidden units is slightly overestimated in this approach. To obtain an estimate of the number of hidden units for a fully connected net with n output units, it is necessary to obtain an estimate of the number of conflicts contained in the individual binary responses that must be learned by each ouput unit. The estimate produced is data dependent, as the number of conflicts for an output unit depends on the specific responses of the output unit to the input vectors contained in the training set.

    Original languageEnglish (US)
    Title of host publicationIJCNN Int Jt Conf Neural Network
    Editors Anon
    Place of PublicationPiscataway, NJ, United States
    PublisherPubl by IEEE
    Pages677-681
    Number of pages5
    StatePublished - 1989
    EventIJCNN International Joint Conference on Neural Networks - Washington, DC, USA
    Duration: Jun 18 1989Jun 22 1989

    Other

    OtherIJCNN International Joint Conference on Neural Networks
    CityWashington, DC, USA
    Period6/18/896/22/89

    Fingerprint

    Neural networks

    ASJC Scopus subject areas

    • Engineering(all)

    Cite this

    Gutierrez, M., Wang, J., & Grondin, R. (1989). Estimating hidden unit number for two-layer perceptrons. In Anon (Ed.), IJCNN Int Jt Conf Neural Network (pp. 677-681). Piscataway, NJ, United States: Publ by IEEE.

    Estimating hidden unit number for two-layer perceptrons. / Gutierrez, Mario; Wang, Jennifer; Grondin, Robert.

    IJCNN Int Jt Conf Neural Network. ed. / Anon. Piscataway, NJ, United States : Publ by IEEE, 1989. p. 677-681.

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    Gutierrez, M, Wang, J & Grondin, R 1989, Estimating hidden unit number for two-layer perceptrons. in Anon (ed.), IJCNN Int Jt Conf Neural Network. Publ by IEEE, Piscataway, NJ, United States, pp. 677-681, IJCNN International Joint Conference on Neural Networks, Washington, DC, USA, 6/18/89.
    Gutierrez M, Wang J, Grondin R. Estimating hidden unit number for two-layer perceptrons. In Anon, editor, IJCNN Int Jt Conf Neural Network. Piscataway, NJ, United States: Publ by IEEE. 1989. p. 677-681
    Gutierrez, Mario ; Wang, Jennifer ; Grondin, Robert. / Estimating hidden unit number for two-layer perceptrons. IJCNN Int Jt Conf Neural Network. editor / Anon. Piscataway, NJ, United States : Publ by IEEE, 1989. pp. 677-681
    @inproceedings{e1c6c7ef5a8a4e0db00b9eec277a9f4a,
    title = "Estimating hidden unit number for two-layer perceptrons",
    abstract = "The authors concentrate on estimating how many hidden units are needed for large nets. The design method is adapted from a more rigorous one developed for smaller nets where all possible input/output relationships are presented in the training vectors. It provides a 'ballpark' estimate of the minimum number of hidden units required by a two-layer perceptron for a provided training subset of the possible input/output pairs. Typically, the number of hidden units is slightly overestimated in this approach. To obtain an estimate of the number of hidden units for a fully connected net with n output units, it is necessary to obtain an estimate of the number of conflicts contained in the individual binary responses that must be learned by each ouput unit. The estimate produced is data dependent, as the number of conflicts for an output unit depends on the specific responses of the output unit to the input vectors contained in the training set.",
    author = "Mario Gutierrez and Jennifer Wang and Robert Grondin",
    year = "1989",
    language = "English (US)",
    pages = "677--681",
    editor = "Anon",
    booktitle = "IJCNN Int Jt Conf Neural Network",
    publisher = "Publ by IEEE",

    }

    TY - GEN

    T1 - Estimating hidden unit number for two-layer perceptrons

    AU - Gutierrez, Mario

    AU - Wang, Jennifer

    AU - Grondin, Robert

    PY - 1989

    Y1 - 1989

    N2 - The authors concentrate on estimating how many hidden units are needed for large nets. The design method is adapted from a more rigorous one developed for smaller nets where all possible input/output relationships are presented in the training vectors. It provides a 'ballpark' estimate of the minimum number of hidden units required by a two-layer perceptron for a provided training subset of the possible input/output pairs. Typically, the number of hidden units is slightly overestimated in this approach. To obtain an estimate of the number of hidden units for a fully connected net with n output units, it is necessary to obtain an estimate of the number of conflicts contained in the individual binary responses that must be learned by each ouput unit. The estimate produced is data dependent, as the number of conflicts for an output unit depends on the specific responses of the output unit to the input vectors contained in the training set.

    AB - The authors concentrate on estimating how many hidden units are needed for large nets. The design method is adapted from a more rigorous one developed for smaller nets where all possible input/output relationships are presented in the training vectors. It provides a 'ballpark' estimate of the minimum number of hidden units required by a two-layer perceptron for a provided training subset of the possible input/output pairs. Typically, the number of hidden units is slightly overestimated in this approach. To obtain an estimate of the number of hidden units for a fully connected net with n output units, it is necessary to obtain an estimate of the number of conflicts contained in the individual binary responses that must be learned by each ouput unit. The estimate produced is data dependent, as the number of conflicts for an output unit depends on the specific responses of the output unit to the input vectors contained in the training set.

    UR - http://www.scopus.com/inward/record.url?scp=0024903275&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=0024903275&partnerID=8YFLogxK

    M3 - Conference contribution

    AN - SCOPUS:0024903275

    SP - 677

    EP - 681

    BT - IJCNN Int Jt Conf Neural Network

    A2 - Anon, null

    PB - Publ by IEEE

    CY - Piscataway, NJ, United States

    ER -