Asynchronous Methods for Deep Reinforcement Learning
This work addresses training stability and efficiency issues for reinforcement learning practitioners, representing a significant but incremental improvement over existing methods.
The authors tackled the problem of unstable training in deep reinforcement learning by proposing an asynchronous framework using parallel actor-learners, which achieved state-of-the-art performance on Atari games while training twice as fast on a CPU instead of a GPU.
We propose a conceptually simple and lightweight framework for deep reinforcement learning that uses asynchronous gradient descent for optimization of deep neural network controllers. We present asynchronous variants of four standard reinforcement learning algorithms and show that parallel actor-learners have a stabilizing effect on training allowing all four methods to successfully train neural network controllers. The best performing method, an asynchronous variant of actor-critic, surpasses the current state-of-the-art on the Atari domain while training for half the time on a single multi-core CPU instead of a GPU. Furthermore, we show that asynchronous actor-critic succeeds on a wide variety of continuous motor control problems as well as on a new task of navigating random 3D mazes using a visual input.