Enhancing Offline Model-Based RL via Active Model Selection: A Bayesian Optimization Perspective
This addresses a key bottleneck in offline RL for researchers and practitioners by enabling more accurate model selection with minimal online interaction, though it is incremental as it builds on existing offline MBRL methods.
The paper tackles the problem of inaccurate model selection in offline model-based reinforcement learning due to distribution shift by proposing BOMS, an active model selection framework using Bayesian optimization, which improves over baseline methods with only 1%-2.5% of online interaction compared to offline data.
Offline model-based reinforcement learning (MBRL) serves as a competitive framework that can learn well-performing policies solely from pre-collected data with the help of learned dynamics models. To fully unleash the power of offline MBRL, model selection plays a pivotal role in determining the dynamics model utilized for downstream policy learning. However, offline MBRL conventionally relies on validation or off-policy evaluation, which are rather inaccurate due to the inherent distribution shift in offline RL. To tackle this, we propose BOMS, an active model selection framework that enhances model selection in offline MBRL with only a small online interaction budget, through the lens of Bayesian optimization (BO). Specifically, we recast model selection as BO and enable probabilistic inference in BOMS by proposing a novel model-induced kernel, which is theoretically grounded and computationally efficient. Through extensive experiments, we show that BOMS improves over the baseline methods with a small amount of online interaction comparable to only $1\%$-$2.5\%$ of offline training data on various RL tasks.