Using reinforcement learning to learn how to play text-based games
This work addresses the challenge of optimizing dialogue systems and similar real-world applications, though it is incremental as it builds on existing reinforcement learning techniques for text games.
The authors tackled the problem of learning optimal control policies in systems with natural language action spaces by using reinforcement learning on text-based games, achieving a general agent capable of playing multiple games simultaneously and introducing an open-source library for future research.
The ability to learn optimal control policies in systems where action space is defined by sentences in natural language would allow many interesting real-world applications such as automatic optimisation of dialogue systems. Text-based games with multiple endings and rewards are a promising platform for this task, since their feedback allows us to employ reinforcement learning techniques to jointly learn text representations and control policies. We present a general text game playing agent, testing its generalisation and transfer learning performance and showing its ability to play multiple games at once. We also present pyfiction, an open-source library for universal access to different text games that could, together with our agent that implements its interface, serve as a baseline for future research.