Learning and Unlearning in Hopfield-Like Neural Network Performing Boolean Factor Analysis

Authors: A. A. Frolov, Dušan Húsek, I. P. Muraviev, P.Y. Polyakov
Publication year: 2010
Publication type: Kapitola v monografii
Periodical: Advances in Machine Learning I
Bibliographic citation: FROLOV, A. A., HÚSEK, Dušan, MURAVIEV, I. P., POLYAKOV, P.Y. Learning and Unlearning in Hopfield-Like Neural Network Performing Boolean Factor Analysis. In: KORONACKI, J., RAS, Z.W., WIERZCHON, S.T., KACPRZYK, J., eds. Advances in Machine Learning I. Berlin: Springer, 2010, S. 501-518. Studies in Computational Intelligence, 262. ISBN 978-3-642-05176-0
Dostupné z: doi:10.1007/978-3-642-05177-7_26
DOI: 10.1007/978-3-642-05177-7_26
ID: 331155

Abstract

The peculiarity of usage the Hopfield-like network for Boolean factor analysis is the appearance of two global spurious attractors. They become dominant and, therefore, prevent successful factors search. To eliminate these attractors we propose a special unlearning procedure. This second unlearning procedure provides the suppression of factors with the largest attraction basins which dominate after suppression of global spurious attractors and prevent the recall of other factors. The origin of the global spurious attractors and the efficiency of the unlearning procedures are investigated in the present paper.