Müller-Schloer, Christian
Sandbox Learning: Try without error?
Abstract
Adaptivity is enabled by learning. Natural systems learn differently from technical systems. In particular, technical systems must not make errors. On the other hand, learning seems to be impossible without occasional errors.
We propose a 3-level architecture for learning in adaptive technical systems and show its applicability in the domains of traffic control and communication network control.
BibTeX - Entry
@InProceedings{mllerschloer:DSP:2009:2123,
author = {Christian M{\"u}ller-Schloer},
title = {Sandbox Learning: Try without error?},
booktitle = {Self-Healing and Self-Adaptive Systems},
year = {2009},
editor = {Artur Andrzejak and Kurt Geihs and Onn Shehory and John Wilkes},
number = {09201},
series = {Dagstuhl Seminar Proceedings},
ISSN = {1862-4405},
publisher = {Schloss Dagstuhl - Leibniz-Zentrum fuer Informatik, Germany},
address = {Dagstuhl, Germany},
URL = {http://drops.dagstuhl.de/opus/volltexte/2009/2123},
annote = {Keywords: Learning, real time, natural systems, technical systems}
}
|
Keywords: |
|
Learning, real time, natural systems, technical systems |
|
Seminar: |
|
09201 - Self-Healing and Self-Adaptive Systems
|
|
Issue date: |
|
2009 |
|
Date of publication: |
|
07.08.2009 |
07.08.2009
