In this paper, we develop a novel learning model that combines directional learning and automata reinforcement. We show that directional learning can be defined on a cyclic order in addition to its original domain (interval) and that pattern learning by reinforcing parts of automata [Chernov,...