Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

DeepDyve requires Javascript to function. Please enable Javascript on your browser to continue.

Qualitative Spatial Abstraction in Reinforcement Learning

Part of the Cognitive Technologies Book Series

DOI: 10.1007/978-3-642-16590-0

ISBN: 978-3-642-16589-4

Publisher: Springer Berlin Heidelberg - Springer Journals

Nov 2, 2010

Available

Introduction

Frommberger, Lutz

Pages: 1 –8

[One of the most essential properties of a cognitive being is its ability to learn. Learningis the “process of acquiring modifications in existing knowledge, skills, habits,or tendencies through experience, practice, or exercise” (Encyclopædia Britannica, 2007).]
Available

Foundations of Reinforcement Learning

Frommberger, Lutz

Pages: 9 –21

[This chapter gives an introduction to the machine learning paradigm of reinforcement learning and introduces basic notations. Following a short overview on machine learning in Sect. 2.1, Sect. 2.2 explains the reinforcement learning model, before the central framework of Markov decision...
Available

Abstraction and Knowledge Transfer in Reinforcement Learning

Frommberger, Lutz

Pages: 23 –41

[This chapter presents the state of the art in research on reinforcement learning with a focus on abstraction and transfer learning. Especially, the open questions of performance in large, continuous state spaces and knowledge transfer are worked out as central challenges of reinforcement...
Available

Qualitative State Space Abstraction

Frommberger, Lutz

Pages: 43 –65

[In this chapter, we take a deeper look into the nature of abstraction. In particular, this chapter argues for the use of state space abstraction (Sect. 4.1) and presents a formal framework of abstraction and its different facets in Sect. 4.2. Based on this framework, a view on the...
Available

Generalization and Transfer Learning with Qualitative Spatial Abstraction

Frommberger, Lutz

Pages: 67 –92

[In this chapter we will investigate the properties of observation space representations achieved by qualitative abstraction with respect to generalization and transfer learning capabilities. Section 5.1 describes the importance of structural similarity for knowledge reuse.]
Available

RLPR – An Aspectualizable State Space Representation

Frommberger, Lutz

Pages: 93 –122

[In this chapter, we get to know a qualitative spatial representation tailored for indoor robot navigation tasks. First, the task to be solved is introduced and task space and structure space are identified along with considerations about frames of reference (Sect. 6.1). In Sect. 6.2 we derive a...
Available

Empirical Evaluation

Frommberger, Lutz

Pages: 123 –154

[This chapter contains an empirical evaluation of the learning performance and generalization and transfer learning capabilities of RLPR-based representations in a goal-directed reinforcement learning task in an indoor office environment. It subsumes experiments performed over a longer period of...
Available

Summary and Outlook

Frommberger, Lutz

Pages: 155 –159

[This chapter concludes this book with a summary and discussion of the presented work and an outlook on future research based on the described results.]

results per page

Prev
1
Next

Browse All eBooks