Learning behavior and temporary minima of two-layer neural networks

Anne J. Annema, Klaas Hoen, Klaas Hoen, Hans Wallinga

    Research output: Contribution to journalArticleAcademicpeer-review

    5 Citations (Scopus)
    81 Downloads (Pure)

    Abstract

    This paper presents a mathematical analysis of the occurrence of temporary minima during training of a single-output, two-layer neural network, with learning according to the back-propagation algorithm. A new vector decomposition method is introduced, which simplifies the mathematical analysis of learning of neural networks considerably. The analysis shows that temporary minima are inherent to multilayer networks learning. A number of numerical results illustrate the analytical conclusions.
    Original languageEnglish
    Pages (from-to)1387-1403
    JournalNeural networks
    Volume7
    Issue number9
    DOIs
    Publication statusPublished - 1994

      Fingerprint

    Keywords

    • IR-15176
    • METIS-112050
    • Temporary minimum
    • Pattern classification
    • Learning
    • Neural Networks
    • Multilayer perceptron
    • Back propagation

    Cite this