… Transfer in reinforcementlearning refers to the notion that generalization should occur not only within a task but also across tasks. We propose a transfer … the reinforcementlearning …
… transfer in reinforcementlearning based on the idea that related tasks share some common features, and that transfer … provides some insight into when transfer can be usefully applied …
ME Taylor, P Stone - … 24th international conference on Machine learning, 2007 - dl.acm.org
… to learn a target task faster. Recently introduced transfer methods in reinforcementlearning settings have shown considerable promise, but they typically transfer between pairs of very …
ME Taylor, G Kuhlmann, P Stone - AAMAS (1), 2008 - cs.utexas.edu
Recent work in transferlearning has succeeded in making reinforcementlearning algorithms more efficient by incorporating knowledge from previous tasks. However, such methods …
… Domain adaptation is an important open problem in deep reinforcementlearning (RL). In many scenarios of interest data is hard to obtain, so agents may learn a source policy in a …
… ReinforcementLearning Background This monograph focuses on transferlearning in reinforcementlearning domains; some RL background is necessary. Our goal in this chapter is to …
ME Taylor, P Stone - Journal of Machine Learning Research, 2009 - jmlr.org
… In this article we present a framework that classifies transferlearning methods in terms of their … Transfer Figure 1: This article focuses on transfer between reinforcementlearning tasks. …
… -of-the-art transferlearning approaches, under which we analyze … reinforcementlearning backbones, and practical applications. We also draw connections between transferlearning …
… learning on the robot. To this end, we introduce a novel reinforcementlearning task for humanoid robots and demonstrate that transferlearning can be effective for this task. The results …