Abstract
This article presents an approximation scheme for the infinite-dimensional linear programming formulation of discrete-time Markov control processes via a finite-dimensional convex program, when the dynamics are unknown and learned from data. We derive a probabilistic explicit error bound between the data-driven finite convex program and the original infinite linear program. We further discuss the sample complexity of the error bound which translates to the number of samples required for an a priori approximation accuracy. Our analysis sheds light on the impact of the choice of basis functions for approximating the true value function. Finally, the relevance of the method is illustrated on a truncated LQG problem.
Original language | English |
---|---|
Title of host publication | Proceedings of the 2017 IEEE 56th Annual Conference on Decision and Control |
Editors | A Astolfi et al |
Place of Publication | Piscataway, NJ, USA |
Publisher | IEEE |
Pages | 5174-5179 |
ISBN (Electronic) | 978-150902873-3 |
DOIs | |
Publication status | Published - 2017 |
Event | CDC 2017: 56th IEEE Annual Conference on Decision and Control - Melbourne, Australia Duration: 12 Dec 2017 → 15 Dec 2017 http://cdc2017.ieeecss.org/ |
Conference
Conference | CDC 2017: 56th IEEE Annual Conference on Decision and Control |
---|---|
Country/Territory | Australia |
City | Melbourne |
Period | 12/12/17 → 15/12/17 |
Other | The CDC is recognized as the premier scientific and engineering conference dedicated to the advancement of the theory and practice of systems and control. The CDC annually brings together an international community of researchers and practitioners in the field of automatic control to discuss new research results, perspectives on future developments, and innovative applications relevant to decision making, systems and control, and related areas. |
Internet address |