Model-Based Reinforcement Learning with Multi-Task Offline Pretraining
This work addresses the problem of efficient reinforcement learning for researchers and practitioners by enabling better transfer from offline pretraining, though it is incremental as it builds on existing model-based and multi-task approaches.
The paper tackles the challenge of transferring knowledge from offline datasets to improve training efficiency in online reinforcement learning tasks by proposing a model-based method that uses world models to measure task relevance for dynamics and policy transfer. It demonstrates advantages over state-of-the-art methods in Meta-World and DeepMind Control Suite benchmarks.
Pretraining reinforcement learning (RL) models on offline datasets is a promising way to improve their training efficiency in online tasks, but challenging due to the inherent mismatch in dynamics and behaviors across various tasks. We present a model-based RL method that learns to transfer potentially useful dynamics and action demonstrations from offline data to a novel task. The main idea is to use the world models not only as simulators for behavior learning but also as tools to measure the task relevance for both dynamics representation transfer and policy transfer. We build a time-varying, domain-selective distillation loss to generate a set of offline-to-online similarity weights. These weights serve two purposes: (i) adaptively transferring the task-agnostic knowledge of physical dynamics to facilitate world model training, and (ii) learning to replay relevant source actions to guide the target policy. We demonstrate the advantages of our approach compared with the state-of-the-art methods in Meta-World and DeepMind Control Suite.