Strictly modular probabilistic neural networks for pattern recognition

0411265 20240903170412.220060210235959.9 Strictly modular probabilistic neural networks for pattern recognition 17 s. cav_un_epca*02903211210-0552Neural Network World136 (2003)599-615Ústav informatiky AV ČR, v. v. i. neural networks distribution mixtures pattern recognition cav_un_auth*0101091 Grim Jiří UTIA-B Department of Pattern Recognition Ústav teorie informace a automatizace AV ČR, v. v. i. cav_un_auth*0213065 Just P. CZ cav_un_auth*0101182 Pudil Pavel UTIA-B Department of Pattern Recognition Ústav teorie informace a automatizace AV ČR, v. v. i. 09K 12B 06D GA402/01/0981 GA ČR cav_un_auth*0008962 CEZ:AV0Z1075907 Considering the statistical pattern recognition we approximate the unknown class-conditional probability distributions by multivariate Bernoulli mixtures. We show that both the parameter optimization based on EM algorithm and the resulting Bayesian decision-making can be realized by a strictly modular probabilistic neural network. The autonomous adaptation of neurons includes only the locally available information. The properties of the sequential learning procedure are illustrated by numerical examples. BB RO http://hdl.handle.net/11104/0003519 UTIA-B 20030252 2003 cav_un_epca*0290321 Neural Network World 1210-0552 Roč. 13 č. 6 2003 599 615 Ústav informatiky AV ČR, v. v. i.