Clapnet-neural net's program simulator for classification and prediction
K. Arutunov, A. Batalov, A. Goryashko, G. Lakshin · 2002
Feedforward networks with intermediate or "hidden" layers between the input and the output layers are considered. The back-propagation algorithm is a main reason for such resurgence of interest in this area because it is very simple and in some cases a wonderfully efficient method of the nets learning. Today there are a lot of program packages for simulation of artificial neural systems. Unfortunately most of them are unavailable for native programmers because of cost, for example, neural net simulator in MATLAB, or functional possibility of these systems is insufficient for developing real-world applications (for example Brain Maker Pro). The authors present a tool kit working under Windows, using above all its graphical interface, to investigate connectionist approaches in the speech processing.