A game-theoretic formulation of adaptive categorization mechanism for ART-type networks is proposed in this paper. We have derived the game-theoretic model ΓAC for competitive processes of categorization of ART-type networks and an update rule for vigilance parameters using the concept of learning automata. Numbers of clusters generated by ART adaptive categorization are similar regardless of the initial vigilance parameters ρ assigned to the ART networks as demonstrated in the experiments provided. The proposed ART adaptive categorization mechanism can thus avoid the problem of choosing suitable vigilance parameter a priori for pattern categorization.
CITATION STYLE
Fung, W. K., & Liu, Y. H. (2001). A game-theoretic adaptive categorization mechanism for ART-type networks. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 2130, pp. 170–176). Springer Verlag. https://doi.org/10.1007/3-540-44668-0_24
Mendeley helps you to discover research relevant for your work.