Natural learning of neural networks by reconfiguration

L. Spaanenburg, R. Alberts, Cornelis H. Slump, B.J. van der Zwaag

    Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

    2 Citations (Scopus)
    105 Downloads (Pure)

    Abstract

    The communicational and computational demands of neural networks are hard to satisfy in a digital technology. Temporal computing addresses this problem by iteration, but leaves a slow network. Spatial computing only became an option with the coming of modern FPGA devices. The paper provides two examples. First the balance between area and time is discussed on the realization of a modular feed-forward network. Second, the design of real-time image processing through a Cellular Neural Network is treated. In both examples, reconfiguration can be applied to provide for a natural and transparent support of learning.
    Original languageUndefined
    Title of host publicationBioengineered and Bioinspired Systems
    EditorsAngel Rodriguez-Vazquez, Derek Abbott, Ricardo Carmona
    Place of PublicationMaspaloma, Spain
    PublisherSPIE
    Pages273-284
    Number of pages12
    ISBN (Print)0-8194-4979-2
    DOIs
    Publication statusPublished - May 2003
    EventBioengineered and Bioinspired Systems - Gran Canaria, Spain
    Duration: 19 May 200321 May 2003

    Publication series

    NameProceedings of SPIE
    PublisherSPIE
    Volume5119

    Conference

    ConferenceBioengineered and Bioinspired Systems
    Period19/05/0321/05/03
    Other19-21 May 2003

    Keywords

    • METIS-211890
    • Spatial Computing
    • Wave Computing
    • Field-Programmable Gate-Array
    • Feed-Forward Neural Network
    • EWI-9668
    • Cellular Neural Network. Temporal computing
    • Reconfiguration
    • Modularity
    • IR-45308

    Cite this