Deep Q-Network

What is a Deep Q-Network (DQN)?

A Deep Q-Network combines Q-learning with deep neural function approximation so agents can learn action-value estimates from high-dimensional inputs such as pixels.

Introduced in deep reinforcement learning research on Atari, DQN popularized experience replay and target networks to stabilize training.

Authoritative reference: Playing Atari with Deep Reinforcement Learning (arXiv)