Massively Parallel Methods for Deep Reinforcement Learning

TLDR

The paper introduces the first massively distributed architecture for deep reinforcement learning. The architecture comprises parallel actors, parallel learners, a distributed neural network, and a distributed experience store, and it was used to implement DQN across 49 Atari 2600 games with identical hyperparameters. The distributed DQN achieved better performance than the non‑distributed version in 41 of 49 Atari games and cut wall‑time by roughly tenfold on most games.

Abstract

We present the first massively distributed architecture for deep reinforcement learning. This architecture uses four main components: parallel actors that generate new behaviour; parallel learners that are trained from stored experience; a distributed neural network to represent the value function or behaviour policy; and a distributed store of experience. We used our architecture to implement the Deep Q-Network algorithm (DQN). Our distributed algorithm was applied to 49 games from Atari 2600 games from the Arcade Learning Environment, using identical hyperparameters. Our performance surpassed non-distributed DQN in 41 of the 49 games and also reduced the wall-time required to achieve these results by an order of magnitude on most games.