Counting to Explore and Generalize in Text-based Games

TL;DR

This paper introduces a recurrent reinforcement learning agent with episodic exploration that effectively learns and generalizes policies in text-based games, outperforming previous approaches especially on unseen, more complex games.

Contribution

The paper presents a novel episodic exploration mechanism integrated into a recurrent RL agent for better generalization in text-based games.

Findings

Agent successfully generalizes to unseen, more difficult games

Achieves promising results on a variety of generated text-based games

Outperforms previous RL approaches in generalization capability

Abstract

We propose a recurrent RL agent with an episodic exploration mechanism that helps discovering good policies in text-based game environments. We show promising results on a set of generated text-based games of varying difficulty where the goal is to collect a coin located at the end of a chain of rooms. In contrast to previous text-based RL approaches, we observe that our agent learns policies that generalize to unseen games of greater difficulty.