Industrial Automatic Control Systems and Controllers Annotation << Back Implementation of a learning Algorithm for Kohonen’s Self-organizing map on Graphical Processor D.P. BOGOLYUBOV, A.A. CHANKIN, K.V. STEMIKOVSKAYA In this article we introduce a CUDA-based implementation of Kohonen self-organizing map. We describe software implementation and test results confirming performance growth with increasing size of neural network comparative to serial version of algorithm. Keywords: SOM, parallel computing, neural networks training algorithm, GPGPU, CUDA Contacts: e-mail: bogolub@mail.ru Pp. 30-35.