This document discusses deep reinforcement learning techniques. It begins with an overview of neural networks as universal function approximators and how they can be used for reinforcement learning agents to choose actions. It then discusses the Deep Q-Network (DQN) specifically, noting that a DQN agent achieved over 75% of the human score in 29 out of 49 Atari games and beat the human score in 22 games by using a neural network to approximate the Q-function.