Reinforcement Learning Technique for Finding the Feedback Capacity