Transfer Learning for Policy Search Methods

Abstract

An ambitious goal of transfer learning is to learn a task faster after training on a different, but related, task. In this paper we extend a previously successful temporal difference (Sutton & Barto, 1998) approach to transfer inreinforcement learning (Sutton & Barto, 1998) tasks to work with policy search. In particular, we show how to construct a mapping to translate a population of policies trained via genetic algorithms (GAs) (Goldberg, 1989) from a source task to a target task. Empirical results in robot soccer Keepaway, a standard RL benchmark domain (Stone et al., 2006), demonstrate that transfer via inter-task mapping can markedly reduce the time required to learn a second, more complex, task.