Ensemble methods for handwritten digit recognition

Lars Kai Hansen, Christian Liisberg, P. Salamon

    Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

    653 Downloads (Pure)


    Neural network ensembles are applied to handwritten digit recognition. The individual networks of the ensemble are combinations of sparse look-up tables (LUTs) with random receptive fields. It is shown that the consensus of a group of networks outperforms the best individual of the ensemble. It is further shown that it is possible to estimate the ensemble performance as well as the learning curve on a medium-size database. In addition the authors present preliminary analysis of experiments on a large database and show that state-of-the-art performance can be obtained using the ensemble approach by optimizing the receptive fields. It is concluded that it is possible to improve performance significantly by introducing moderate-size ensembles; in particular, a 20-25% improvement has been found. The ensemble random LUTs, when trained on a medium-size database, reach a performance (without rejects) of 94% correct classification on digits written by an independent group of people
    Original languageEnglish
    Title of host publicationProceedings of the IEEE-SP Workshop Neural Networks for Signal Processing
    Publication date1992
    ISBN (Print)0-7803-0557-4
    Publication statusPublished - 1992
    Event1992 IEEE Workshop on Neural Networks for Signal Processing - Hotel Marielyst, Helsingoer, Denmark
    Duration: 31 Aug 19922 Sept 1992


    Conference1992 IEEE Workshop on Neural Networks for Signal Processing
    LocationHotel Marielyst
    Internet address

    Bibliographical note

    Copyright: 1992 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE


    Dive into the research topics of 'Ensemble methods for handwritten digit recognition'. Together they form a unique fingerprint.

    Cite this