PROBABILISTIC SELF-ORGANIZING MAP AND RADIAL BASIS FUNCTION NETWORKS

We propose in this paper a new learning algorithm probabilistic self-o rganizing map (PRSOM) using a probabilistic formalism for topological maps. This algorithm approximates the density distribution of the inpu t set with a mixture of normal distributions. The unsupervised learnin g is based on the dynamic clusters principle and optimizes the likelih ood function. A supervised version of this algorithm based on radial b asis functions (RBF) is proposed. In order to validate the theoretical approach, we achieve regression tasks on simulated and real data usin g the PRSOM algorithm. Moreover, our results are compared with normali zed Gaussian basis functions (NGBF) algorithm. (C) 1998 Published by E lsevier Science B.V. All rights reserved.