Transfer in Reinforcement Learning via Shared Features
Name
Konidaris-2012-Transfer in Reinforcement Learning via Shared Features.pdf
Size
424.05 KB
Format
Adobe PDF
Checksum (MD5)
930e29f681206e29665e3c50e9f7fc67
Author(s)
Konidaris, George
Scheidwasser, Ilya
Barto, Andrew G.
Date Issued
June 2012
Journal
Journal of Machine Learning Research
Publisher
Journal of Machine Learning Research
Citation
George Konidaris, Ilya Scheidwasser, and Andrew Barto. 2012. Transfer in Reinforcement Learning via Shared Features. Journal of Machine Learning Research 98888 (June 2012), 1333-1371.
Version
Final published version
Abstract
We present a framework for transfer in reinforcement learning based on the idea that related tasks share some common features, and that transfer can be achieved via those shared features. The framework attempts to capture the notion of tasks that are related but distinct, and provides some insight into when transfer can be usefully applied to a problem sequence and when it cannot. We apply the framework to the knowledge transfer problem, and show that an agent can learn a portable shaping function from experience in a sequence of tasks to significantly improve performance in a later related task, even given a very brief training period. We also apply the framework to skill transfer, to show that agents can learn portable skills across a sequence of tasks that significantly improve performance on later related tasks, approaching the performance of agents given perfectly learned problem-specific skills.
MIT Department
Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory
Terms of Use
Article is made available in accordance with the publisher's policy and may be subject to US copyright law. Please refer to the publisher's site for terms of use.
Persistent DSpace Link
DOI of Published Version
http://dl.acm.org/citation.cfm?id=2343689
