Snytkin M. Image processing by neural network ART

The analysis of classification, clustering and image processing problems with using of ANN ART was performed in the thesis. ART neural network structure was investigated. Basic ART network functioning principles were examined. The present algorithms of neural network functioning and training were examined. The network properties analysis during selecting different values of training parameters were investigated. The problem of ART network type selection for particular solution was examined. Effective strategies and methods of ART neural network training for classification and clustering problem solution were developed. The method based on neural network ART which was used for vector quantization in image compression scheme was developed. The simulation of different types of neural network ART was performed. The process of handling tasks of classification, clustering and image compression was investigated. The authenticity of results is confirmed by experimental researches and applications.