reinforcement_learning_bht