ML RL with neural network on Asteroids Game
I have built a DQ network ,which analyses the states of asteroids and the rocket gives out the best possible action to be performed (left, right ,shoot , idle).But the agent keeps spamming bullets ?
can anyone help me in solving this problem ?