Relative reward strength algorithms for learning automata | IEEE Journals & Magazine | IEEE Xplore