elib
DLR-Header
DLR-Logo -> http://www.dlr.de
DLR Portal Home | Imprint | Contact | Deutsch
Fontsize: [-] Text [+]

Sequential feature selection for classification

Rückstieß, Thomas and Osendorfer, Christian and van der Smagt, Patrick (2011) Sequential feature selection for classification. In: Proceedings. ICML 2011, The 28th International Conference on Machine Learning, 05.– 08.Dez. 2011, Perth, Australia.

[img] PDF - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader
402kB

Abstract

In most real-world information processing problems, data is not a free resource; its acquisition is rather time-consuming and/or expensive. We investigate how these two factors can be included in supervised classi cation tasks by deriving classi cation as a sequential decision process and making it accessible to Reinforcement Learning. Our method performs a sequential feature selection that learns which features are most informative at each timestep, choosing the next feature depending on the already selected features and the internal belief of the classi er. Experiments on a handwritten digits classi cation task show signi cant reduction in required data for correct classi cation, while a medical diabetes prediction task illustrates variable feature cost minimization as a further property of our algorithm.

Document Type:Conference or Workshop Item (Speech, Paper)
Title:Sequential feature selection for classification
Authors:
AuthorsInstitution or Email of Authors
Rückstieß, Thomas Technische Universität München
Osendorfer, Christian Technische Universität München
van der Smagt, Patrick smagt@dlr.de
Date:2011
Journal or Publication Title:Proceedings
Refereed publication:Yes
In ISI Web of Science:No
Status:Published
Keywords:reinforcement learning, feature selection, classi cation
Event Title:ICML 2011, The 28th International Conference on Machine Learning
Event Location:Perth, Australia
Event Type:international Conference
Event Dates:05.– 08.Dez. 2011
HGF - Research field:Aeronautics, Space and Transport (old)
HGF - Program:Space (old)
HGF - Program Themes:W SY - Technik für Raumfahrtsysteme
DLR - Research area:Space
DLR - Program:W SY - Technik für Raumfahrtsysteme
DLR - Research theme (Project):W - RMC - Kognitive Intelligenz und Autonomie (old)
Location: Oberpfaffenhofen
Institutes and Institutions:Institute of Robotics and Mechatronics > Robotic Systems
Deposited By: Gabriele Beinhofer
Deposited On:20 Jan 2012 11:32
Last Modified:20 Jan 2012 11:32

Repository Staff Only: item control page

Browse
Search
Help & Contact
Informationen
electronic library is running on EPrints 3.3.12
Copyright © 2008-2012 German Aerospace Center (DLR). All rights reserved.