File(s) under permanent embargo

Improving reinforcement learning with interactive feedback and affordances

conference contribution
posted on 01.01.2014, 00:00 authored by F Cruz Naranjo, S Magg, C Weber, S Wermter
Interactive reinforcement learning constitutes an alternative for improving convergence speed in reinforcement learning methods. In this work, we investigate inter-agent training and present an approach for knowledge transfer in a domestic scenario where a first agent is trained by reinforcement learning and afterwards transfers selected knowledge to a second agent by instructions to achieve more efficient training. We combine this approach with action-space pruning by using knowledge on affordances and show that it significantly improves convergence speed in both classic and interactive reinforcement learning scenarios.

History

Event

European Society for Cognitive Systems. Conference (4th : 2014 : Genoa, Italy)

Series

European Society for Cognitive Systems Conference

Pagination

165 - 170

Publisher

Institute of Electrical and Electronics Engineers

Location

Genoa, Italy

Place of publication

Piscataway, N.J.

Start date

13/10/2014

End date

16/10/2014

ISBN-13

9781479975402

Language

eng

Publication classification

E1.1 Full written paper - refereed

Copyright notice

2014, IEEE

Editor/Contributor(s)

[Unknown]

Title of proceedings

IEEE ICDL-EPIROB 2014 : Proceedings of the 4th Joint IEEE International Conference on Development and Learning and on Epigenetic Robotics 2014