AlphaGo and AlphaZero both rely on reinforcement learning to train. They also use deep neural networks as part of the reinforcement learning network, to predict outcome probabilities. In this ...