Reinforcement-Learning

Deep Recurrent Q learning

• Take successive image as Input of Network
• By Reccurent Network, Agent can predict Ball moving and move to the direction to block the ball

Structure

• Image -> CNN -> RNN -> Q value