Beating Atari with Natural Language Guided Reinforcement Learning
Artificial Intelligence
2017-04-20 v1
Abstract
We introduce the first deep reinforcement learning agent that learns to beat Atari games with the aid of natural language instructions. The agent uses a multimodal embedding between environment observations and natural language to self-monitor progress through a list of English instructions, granting itself reward for completing instructions in addition to increasing the game score. Our agent significantly outperforms Deep Q-Networks (DQNs), Asynchronous Advantage Actor-Critic (A3C) agents, and the best agents posted to OpenAI Gym on what is often considered the hardest Atari 2600 environment: Montezuma's Revenge.
Keywords
Cite
@article{arxiv.1704.05539,
title = {Beating Atari with Natural Language Guided Reinforcement Learning},
author = {Russell Kaplan and Christopher Sauer and Alexander Sosa},
journal= {arXiv preprint arXiv:1704.05539},
year = {2017}
}