0

Bootstrapped Q-learning with Context Relevant Observation Pruning to Generalize in Text-based Games

CREST, a method for removing irrelevant tokens in text-based games, improves RL model generalization by reducing overfitting and decreasing training game requirements.

Year
2020
Venue
EMNLP 2020 11
Authors
6
Hosting
Abstract onlyARXIV-DEFAULT

Cite

Notes

Only stored in your browser.

Attribution

Abstract & full text
arxiv.org/abs/2009.11896ARXIV-DEFAULT
TL;DR
Semantic Scholar
Attribution policy →

Abstract

We show that Reinforcement Learning (RL) methods for solving Text-Based Games (TBGs) often fail to generalize on unseen games, especially in small data regimes. To address this issue, we propose Context Relevant Episodic State Truncation (CREST) for irrelevant token removal in observation text for improved generalization. Our method first trains a base model using Q-learning, which typically overfits the training games. The base model's action token distribution is used to perform observation pruning that removes irrelevant tokens. A second bootstrapped model is then retrained on the pruned observation text. Our bootstrapped agent shows improved generalization in solving unseen TextWorld games, using 10x-20x fewer training games compared to previous state-of-the-art methods despite requiring less number of training episodes.

Authors

6