Modèles de cartes auto-organisatrices hiérarchiques pour la classification phonémique

Abstract

This paper presents hierarchical self-organizing map (som) models for phoneme classification. The hierarchical som method uses a non supervised learning and a spatial organization of data. This classification approach extends the Kohonen map by introducing the principle of multiple prototype vectors by means of an enrichment auxiliary information method in a map. The case study of hierarchical som classification models is phoneme recognition in continuous speech and speaker independent context. The proposed som models serve as tools for developing intelligent systems and pursuing artificial intelligence applications.