Reinforcement Learning

Reinforcement learning Basics

Reinforcement learning - optimal control, where agent learns to develop an optimal policy of sequential actions to take by interacting with an environment. There are various branches in RL - temporal difference, Monte Carlo and dynamic programming.

Deep Learning is more of complex function approximation for image recognition (CNN), speech (RNN) as well dimension reduction and deep network pretraining (unsupervised).

How deep learning and reinforcement learning combines: eg. deep Q learning, Google DeepMind Atari. Deep neural network is used to approximate the Q function in Q-learning (part of temporal difference learning)

1. Deep Q Net Paper
2. Reinforcement Learning with TensorFlow
3. Deep Reinforcement Learning: Pong From Pixel