Hands-On Neuroevolution with Python

Deep Neuroevolution

In this chapter, we presented you with the concept of deep neuroevolution, which can be used to train Deep Artificial Neural Networks (DNNs). You learned how deep neuroevolution can be used to train Atari game-playing agents using the deep reinforcement learning algorithm.

We started with a discussion of the basic concepts behind reinforcement learning. We paid special attention to the popular Q-learning algorithm, which is one of the classic implementations of reinforcement learning. After that, you learned how a DNN could be used to approximate the Q-value function for complex tasks that cannot be approximated by a simple action-state table with Q-values. Next, we discussed how the neuroevolution-based method could be ...