Abstract | ||
---|---|---|
The ability to combine known skills to create new ones may be crucial in the solution of complex reinforcement learning problems that unfold over extended periods. We argue that a robust way of combining skills is to define and manipulate them in the space of pseudo-rewards (or "cumulants"). Based on this premise, we propose a framework for combining skills using the formalism of options. We show that every deterministic option can be unambiguously represented as a cumulant defined in an extended domain. Building on this insight and on previous results on transfer learning, we show how to approximate options whose cumulants are linear combinations of the cumulants of known options. This means that, once we have learned options associated with a set of cumulants, we can instantaneously synthesise options induced by any linear combination of them, without any learning involved. We describe how this framework provides a hierarchical interface to the environment whose abstract actions correspond to combinations of basic skills. We demonstrate the practical benefits of our approach in a resource management problem and a navigation task involving a quadrupedal simulated robot. |
Year | Venue | Keywords |
---|---|---|
2019 | ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 32 (NIPS 2019) | reinforcement learning,transfer learning,basic skills,linear combinations |
Field | DocType | Volume |
Computer science,Human–computer interaction,Artificial intelligence,Machine learning,Reinforcement learning | Conference | 32 |
ISSN | Citations | PageRank |
1049-5258 | 1 | 0.36 |
References | Authors | |
0 | 11 |
Name | Order | Citations | PageRank |
---|---|---|---|
André Barreto | 1 | 12 | 5.65 |
diana borsa | 2 | 11 | 5.00 |
Hou, Shaobo | 3 | 1 | 0.70 |
Gheorghe Comanici | 4 | 1 | 1.37 |
Aygün, Eser | 5 | 1 | 0.36 |
Hamel, Philippe | 6 | 1 | 0.36 |
Toyama, Daniel | 7 | 1 | 0.36 |
Jonathan J Hunt | 8 | 478 | 21.61 |
Shibl Mourad | 9 | 9 | 1.05 |
David Silver | 10 | 8252 | 363.86 |
Doina Precup | 11 | 2829 | 221.83 |