Continuous-time Model-based Reinforcement Learning
Cagatay Yildiz, Markus Heinonen, Harri Lähdesmäki
Abstract
Model-based reinforcement learning (MBRL) approaches rely on discrete-time state transition models whereas physical systems and the vast majority of control tasks operate in continuous-time. To avoid time-discretization approximation of the underlying process, we propose a continuous-time MBRL framework based on a novel actor-critic method. Our approach also infers the unknown state evolution differentials with Bayesian neural ordinary differential equations (ODE) to account for epistemic uncertainty. We implement and test our method on a new ODE-RL suite that explicitly solves continuous-time control systems. Our experiments illustrate that the model is robust against irregular and noisy data, and can solve classic control problems in a sample-efficient manner.
BibTeX
@InProceedings{pmlr-v139-yildiz21a,
title = {Continuous-time Model-based Reinforcement Learning},
author = {Yildiz, Cagatay and Heinonen, Markus and L{\"a}hdesm{\"a}ki, Harri},
booktitle = {Proceedings of the 38th International Conference on Machine Learning},
pages = {12009--12018},
year = {2021},
editor = {Meila, Marina and Zhang, Tong},
volume = {139},
series = {Proceedings of Machine Learning Research},
month = {18--24 Jul},
publisher = {PMLR},
pdf = {http://proceedings.mlr.press/v139/yildiz21a/yildiz21a.pdf},
url = {https://proceedings.mlr.press/v139/yildiz21a.html},
abstract = {Model-based reinforcement learning (MBRL) approaches rely on discrete-time state transition models whereas physical systems and the vast majority of control tasks operate in continuous-time. To avoid time-discretization approximation of the underlying process, we propose a continuous-time MBRL framework based on a novel actor-critic method. Our approach also infers the unknown state evolution differentials with Bayesian neural ordinary differential equations (ODE) to account for epistemic uncertainty. We implement and test our method on a new ODE-RL suite that explicitly solves continuous-time control systems. Our experiments illustrate that the model is robust against irregular and noisy data, and can solve classic control problems in a sample-efficient manner.}
}