We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: Physics-informed Dyna-Style Model-Based Deep Reinforcement Learning for Dynamic Control

Abstract: Model-based reinforcement learning (MBRL) is believed to have much higher sample efficiency compared to model-free algorithms by learning a predictive model of the environment. However, the performance of MBRL highly relies on the quality of the learned model, which is usually built in a black-box manner and may have poor predictive accuracy outside of the data distribution. The deficiencies of the learned model may prevent the policy from being fully optimized. Although some uncertainty analysis-based remedies have been proposed to alleviate this issue, model bias still poses a great challenge for MBRL. In this work, we propose to leverage the prior knowledge of underlying physics of the environment, where the governing laws are (partially) known. In particular, we developed a physics-informed MBRL framework, where governing equations and physical constraints are utilized to inform the model learning and policy search. By incorporating the prior information of the environment, the quality of the learned model can be notably improved, while the required interactions with the environment are significantly reduced, leading to better sample efficiency and learning performance. The effectiveness and merit have been demonstrated over a handful of classic control problems, where the environments are governed by canonical ordinary/partial differential equations.
Comments: 23 figures; 2 tables
Subjects: Machine Learning (cs.LG); Optimization and Control (math.OC)
DOI: 10.1098/rspa.2021.0618
Cite as: arXiv:2108.00128 [cs.LG]
  (or arXiv:2108.00128v1 [cs.LG] for this version)

Submission history

From: Jian-Xun Wang [view email]
[v1] Sat, 31 Jul 2021 02:19:36 GMT (1836kb,D)

Link back to: arXiv, form interface, contact.