Efficient Reinforcement Learning using Gaussian Processes
Abstract:
This book examines Gaussian processes in both model-based reinforcement learning (RL) and inference in nonlinear dynamic systems.
First, we introduce PILCO, a fully Bayesian approach for efficient RL in continuous-valued state and action spaces when no expert knowledge is available. PILCO takes model uncertainties consistently into account during long-term planning to reduce model bias.
Second, we propose principled algorithms for robust filtering and smoothing in GP dynamic systems.
First, we introduce PILCO, a fully Bayesian approach for efficient RL in continuous-valued state and action spaces when no expert knowledge is available. PILCO takes model uncertainties consistently into account during long-term planning to reduce model bias.
Second, we propose principled algorithms for robust filtering and smoothing in GP dynamic systems.
| Zugehörige Institution(en) am KIT | Fakultät für Informatik – Institut für Anthropomatik (IFA) |
| Publikationstyp | Hochschulschrift |
| Publikationsjahr | 2010 |
| Sprache | Englisch |
| Identifikator | ISBN: 978-3-86644-569-7 ISSN: 1867-3813 urn:nbn:de:0072-197996 KITopen-ID: 1000019799 |
| Verlag | KIT Scientific Publishing |
| Umfang | IX, 205 S. |
| Serie | Karlsruhe Series on Intelligent Sensor-Actuator-Systems / Karlsruher Institut für Technologie, Intelligent Sensor-Actuator-Systems Laboratory ; 9 |
| Art der Arbeit | Dissertation |
| Fakultät | Fakultät für Informatik (INFORMATIK) |
| Institut | Fakultät für Informatik – Institut für Anthropomatik (IFA) |
| Prüfungsdaten | 04.12.2009 |
| Schlagwörter | machine learning, Gaussian processes, Bayesian inference, autonomous learning, control |
| Referent/Betreuer | Hanebeck, U. |