Model-based micro-data reinforcement learning: what are the crucial model properties and which model to choose?
This work addresses model selection for micro-data reinforcement learning, offering practical insights for researchers and practitioners, though it is incremental in refining existing methods.
The paper tackled the problem of selecting generative models for model-based reinforcement learning with limited data, finding that mixture density networks excel in multimodal environments while deterministic models perform comparably to probabilistic ones when multimodality is not required, and improved sample complexity on Acrobot by two to four folds.
We contribute to micro-data model-based reinforcement learning (MBRL) by rigorously comparing popular generative models using a fixed (random shooting) control agent. We find that on an environment that requires multimodal posterior predictives, mixture density nets outperform all other models by a large margin. When multimodality is not required, our surprising finding is that we do not need probabilistic posterior predictives: deterministic models are on par, in fact they consistently (although non-significantly) outperform their probabilistic counterparts. We also found that heteroscedasticity at training time, perhaps acting as a regularizer, improves predictions at longer horizons. At the methodological side, we design metrics and an experimental protocol which can be used to evaluate the various models, predicting their asymptotic performance when using them on the control problem. Using this framework, we improve the state-of-the-art sample complexity of MBRL on Acrobot by two to four folds, using an aggressive training schedule which is outside of the hyperparameter interval usually considered