Title :
Design of nonlinear stochastic learning automata
Author :
Neville, R.G. ; Nicol, C.R. ; Mars, P.
Author_Institution :
Robert Gordon´´s Institute of Technology, School of Electronic and Electrical Engineering, Aberdeen, UK
Abstract :
The hardware design of nonlinear stochastic learning automata using adpative digital logic elements is considered. Such techniques, based on digital stochastic computing, are shown to provide faster convergence rates than automata based on linear learning algorithms. Experimental results are presented and validation obtained for theoretical predictions concerning optimal convergence.
Keywords :
learning systems; logic design; special purpose computers; stochastic automata; adaptive digital logic elements; convergence rates; digital stochastic computing; hardware design; nonlinear stochastic learning automata; optimal convergence;
Journal_Title :
Electronics Letters
DOI :
10.1049/el:19780267
