Structure Detection for Contextual Reinforcement Learning

TL;DR

This paper introduces a dynamic framework for identifying problem structures in Contextual Reinforcement Learning, enabling adaptive task transfer strategies that improve performance across diverse environments.

Contribution

The work presents SD-MBTL, a novel framework that detects underlying structures in CMDPs and adaptively switches between transfer learning algorithms for better generalization.

Findings

M/GP-MBTL outperforms prior methods by 12.49% on aggregated metrics.

The framework effectively identifies structures like Mountain in CRL environments.

Experiments on synthetic and real benchmarks validate the adaptive approach's superiority.

Abstract

Contextual Reinforcement Learning (CRL) tackles the problem of solving a set of related Contextual Markov Decision Processes (CMDPs) that vary across different context variables. Traditional approaches--independent training and multi-task learning--struggle with either excessive computational costs or negative transfer. A recently proposed multi-policy approach, Model-Based Transfer Learning (MBTL), has demonstrated effectiveness by strategically selecting a few tasks to train and zero-shot transfer. However, CMDPs encompass a wide range of problems, exhibiting structural properties that vary from problem to problem. As such, different task selection strategies are suitable for different CMDPs. In this work, we introduce Structure Detection MBTL (SD-MBTL), a generic framework that dynamically identifies the underlying generalization structure of CMDP and selects an appropriate MBTL algorithm. For instance, we observe Mountain structure in which generalization performance degrades from the training performance of the target task as the context difference increases. We thus propose M/GP-MBTL, which detects the structure and adaptively switches between a Gaussian Process-based approach and a clustering-based approach. Extensive experiments on synthetic data and CRL benchmarks--covering continuous control, traffic control, and agricultural management--show that M/GP-MBTL surpasses the strongest prior method by 12.49% on the aggregated metric. These results highlight the promise of online structure detection for guiding source task selection in complex CRL environments.