keras实现REINFORCE算法强化学习： # Policy Gradient Minimal implementation of Stochastic Policy Gradient Algorithm in Keras ## Pong Agent  This PG agent seems to get more frequent wins after about 8000 episodes. Below is the score graph.