Learning to Generalize for Sequential Decision Making

Abstract

We consider problems of making sequences of decisions to accomplish tasks, interacting via the medium of language. These problems are often tackled with reinforcement learning approaches. We find that these models do not generalize well when applied to novel task domains. However, the large amount of computation necessary to adequately train and explore the search space of sequential decision making, under a reinforcement learning paradigm, precludes the inclusion of large contextualized language models, which might otherwise enable the desired generalization ability. We introduce a teacher-student imitation learning methodology and a means of converting a reinforcement learning model into a natural language understanding model. Together, these methodologies enable the introduction of contextualized language models into the sequential decision making problem space. We show that models can learn faster and generalize more, leveraging both the imitation learning and the reformulation. Our models exceed teacher performance on various held-out decision problems, by up to 7% on in-domain problems and 24% on out-of-domain problems.

arXiv commentFindings of EMNLP2020, 18 pages

References in corpus (6)

• Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge
• TransferTransfo: A Transfer Learning Approach for Neural Network Based Conversational Agents
• TextWorld: A Learning Environment for Text-based Games
• Counting to Explore and Generalize in Text-based Games
• Action Assembly: Sparse Imitation Learning for Text Based Games with Combinatorial Action Spaces
• Language Expansion In Text-Based Games