On the Expressivity of Neural Networks for Deep Reinforcement Learning
Kefan Dong, Yuping Luo, Tianhe Yu, Chelsea Finn, Tengyu Ma
Abstract
We compare the model-free reinforcement learning with the model-based approaches through the lens of the expressive power of neural networks for policies, Q-functions, and dynamics. We show, theoretically and empirically, that even for one-dimensional continuous state space, there are many MDPs whose optimal Q-functions and policies are much more complex than the dynamics. For these MDPs, model-based planning is a favorable algorithm, because the resulting policies can approximate the optimal policy significantly better than a neural network parameterization can, and model-free or model-based policy optimization rely on policy parameterization. Motivated by the theory, we apply a simple multi-step model-based bootstrapping planner (BOOTS) to bootstrap a weak Q-function into a stronger policy. Empirical results show that applying BOOTS on top of model-based or model-free policy optimization algorithms at the test time improves the performance on benchmark tasks.
BibTeX
@InProceedings{pmlr-v119-dong20d,
title = {On the Expressivity of Neural Networks for Deep Reinforcement Learning},
author = {Dong, Kefan and Luo, Yuping and Yu, Tianhe and Finn, Chelsea and Ma, Tengyu},
booktitle = {Proceedings of the 37th International Conference on Machine Learning},
pages = {2627--2637},
year = {2020},
editor = {III, Hal Daumé and Singh, Aarti},
volume = {119},
series = {Proceedings of Machine Learning Research},
month = {13--18 Jul},
publisher = {PMLR},
pdf = {http://proceedings.mlr.press/v119/dong20d/dong20d.pdf},
url = {https://proceedings.mlr.press/v119/dong20d.html},
abstract = {We compare the model-free reinforcement learning with the model-based approaches through the lens of the expressive power of neural networks for policies, Q-functions, and dynamics. We show, theoretically and empirically, that even for one-dimensional continuous state space, there are many MDPs whose optimal Q-functions and policies are much more complex than the dynamics. For these MDPs, model-based planning is a favorable algorithm, because the resulting policies can approximate the optimal policy significantly better than a neural network parameterization can, and model-free or model-based policy optimization rely on policy parameterization. Motivated by the theory, we apply a simple multi-step model-based bootstrapping planner (BOOTS) to bootstrap a weak Q-function into a stronger policy. Empirical results show that applying BOOTS on top of model-based or model-free policy optimization algorithms at the test time improves the performance on benchmark tasks.}
}