Neural Q-learning

Abstract

In this paper we introduce a novel neural reinforcement learning method. Unlike existing methods, our approach does not need a model of the system and can be trained directly using the measurements of the system. We achieve this by only using one function approximator and approximate the improved policy from this. An experiment using a mobile robot shows that it can be trained using a real system within reasonable time.

Similar content being viewed by others

Use of Neural Networks in Q-Learning Algorithm

Chapter © 2018

Deep Q-Networks

Chapter © 2020

Research on Robot Control Based on Reinforcement Learning

Chapter © 2020

Explore related subjects

Discover the latest articles, news and stories from top researchers in related subjects.

• Artificial Intelligence

Author information

Authors and Affiliations

1. Faculty of Science, University of Amsterdam, Kruislaan 403, 1098SJ, Amsterdam

   Stephan ten Hagen & Ben Kröse

Authors

1. Stephan ten Hagen
   View author publications

   You can also search for this author in PubMed Google Scholar

2. Ben Kröse
   View author publications

   You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Stephan ten Hagen.

Rights and permissions

Reprints and permissions

About this article

Cite this article

ten Hagen, S., Kröse, B. Neural Q-learning. Neural Comput&Applic 12, 81–88 (2003). https://doi.org/10.1007/s00521-003-0369-9

Download citation

• Received: 17 October 2002

• Accepted: 09 April 2003

• Issue Date: November 2003

• DOI: https://doi.org/10.1007/s00521-003-0369-9

Keywords

• Feed-forward network
• Learning from real systems
• Nonlinear systems
• Optimal control Reinforcement learning