Deep Reinforcement Learning with a Natural Language Action Space

Ji He, Jianshu Chen, Xiaodong He +4

This paper introduces a novel architecture for reinforcement learning with deep neural networks designed to handle state and action spaces characterized by natural language, as fou…