Playing Atari with Deep Reinforcement Learning
The Deep Q-Network learned Atari games at superhuman level from raw pixels with experience replay — deep reinforcement learning's founding result.
4 reports · 2013–2017
Agents that learn from reward — value-based methods, policy gradients and self-play.
The Deep Q-Network learned Atari games at superhuman level from raw pixels with experience replay — deep reinforcement learning's founding result.
A3C trains deep RL agents with parallel asynchronous workers, beating DQN on Atari in a fraction of the time on a single CPU.
AlphaGo Zero mastered Go from pure self-play with no human games, beating the version that beat Lee Sedol — a milestone for reinforcement learning.
PPO stabilizes policy gradients with a clipped objective, becoming the default RL algorithm behind robotics, games and RLHF fine-tuning.