Despite early successes, however, reinforcement learning and related work on artificial neural networks fell out of favor and was for years overshadowed by efforts to build AI using symbols and ...