Optimization Of Ann-Training Geometric Algorithm In Independent Component Analysis (Ica)
Yakubovsky Dmitry, Bormontov Evgeny Nikolaevich, Klukin Vladimir Ivanovich, Turikov Dmitry Aleksandrovich · Zenodo (CERN European Organization for Nuclear Research) · 2014
One of the most actual tasks of information processing today is the problem of separating a mixture of several sources of signals. Moreover, the solution requires the on-line processing mode. With this task successfully cope artificial neural networks (ANN). But for stable operation of the system, the ANN require training on specific algorithms. This article discusses a modification one of ANN-training geometric algorithms - histGEO.