This thesis develops methods of task localization, task similarity discovery, and task transfer for eventual use in a reinforcement learning task library system, which can effectively “learn to learn,” improving its performance as it encounters various tasks over the lifetime of the learning system.
College and Department
Physical and Mathematical Sciences; Computer Science
BYU ScholarsArchive Citation
Carroll, James Lamond, "Task Localization, Similarity, and Transfer; Towards a Reinforcement Learning Task Library System" (2005). All Theses and Dissertations. 579.
Reinforcement Learning, Learning to Learn, Lifelong Learning, Task Transfer, Task Libraries