Introduction
Transfer
learning refers to the process of conveying experience from a simple
task to another more complex (and related) task in order to reduce the
amount of time that is required to learn the latter task. Typically,
in a transfer learning procedure the agent learns a behavior in
a source task, and it uses the gained knowledge in order to
speed up the learning process in a \emph{target task}. Reinforcement
Learning algorithms are time expensive
when they learn from scratch, especially in complex domains, and
transfer learning comprises a suitable solution
to speed up the training process.
Bibliography
Source Code
Here you can find our transfer learning
algorithm for the
Mountain Car Domain.
The software is distributed under the GNU
GPL licence. It requires RL-Glue
3.0 and tile
coding. Please
contact Ioannis
Partalas for bug reports, comments, suggestions or request
for help with the source code.
Source code developers: Ioannis Partalas.
Publications
- A. Fachantidis, I. Partalas, G. Tsoumakas, I. Vlahavas, “Transferring Models in Hybrid Reinforcement Learning Agents”, Accepted for presentation at the 12th Conference on Engineering Application of Neural Networks and to be published in the Conference Proceedings, 2011.
- I. Partalas, G. Tsoumakas, K. Tzevanidis,
I. Vlahavas, "Transerring Experience in Reinforcement Learning through
Task Decomposition", Proc. 8th International Conferences on Autonomous
Agents and Multiagent Systems (AAMAS 2009), Budapest, Hungary,2009.
|
