PARALLEL ALGORITHMS FOR MODULES OF LEARNING AUTOMATA

Parallel algorithms are presented for modules of learning automata wit h the objective of improving their speed of convergence without compro mising accuracy. A general procedure suitable for parallelizing a larg e class of sequential learning algorithms on a shared memory system is proposed. Results are derived to shea the quantitative improvements i n speed obtainable using parallelization, The efficacy of the procedur e is demonstrated by simulation studies on algorithms for common payof f games, parametrized learning automata and pattern classification pro blems with noisy classification of training samples.