Model-based micro-data reinforcement learning: what are the crucial model properties and which model to choose?

Balázs Kégl, Gabriel Hurtado, Albert Thomas

Keywords: model-based reinforcement learning, generative models, mixture density nets, dynamic systems, heteroscedasticity

Abstract: 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.

One-sentence Summary: Crucial model properties for model-based reinforcement learning: multi-modal posterior predictives and heteroscedasticity.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

Code: [ ramp-kits/rl_simulator](https://github.com/ramp-kits/rl_simulator)