Wrapped feature selection by means of guided neural network optimisation
Wrapped feature selection by means of guided neural network optimisation
In this paper, we discuss the implementation of a wrapped neural network feature selection approach, introduced here as the Weight Cascaded Retraining (WCR) algorithm. The first part of the paper provides an outline of the algorithm and elaborates on its formal underpinnings. Central to the whole feature pruning approach is the iteratively conceived guided function optimization realized by passing the optimized weight vector from one iteration step to the next. This essentially gives rise to a cascaded form of neural network retraining. In the second part of the paper the theoretical exposition of the WCR algorithm will be illuminated and benchmarked by means of publicly available UCI case material. It is illustrated that WCR based neural network feature selection may be very effective in reducing model complexity for classification modeling via neural networks.
Baesens, B.
f7c6496b-aa7f-4026-8616-ca61d9e216f0
Viaene, S.
68e01ebc-8a4d-4460-86bf-18711fcee8d6
Vanthienen, J.
808131f1-b77b-4dee-bdda-90a94124c999
Dedene, G.
8afb894b-10b1-48d9-8751-4fab7a71b8ca
2000
Baesens, B.
f7c6496b-aa7f-4026-8616-ca61d9e216f0
Viaene, S.
68e01ebc-8a4d-4460-86bf-18711fcee8d6
Vanthienen, J.
808131f1-b77b-4dee-bdda-90a94124c999
Dedene, G.
8afb894b-10b1-48d9-8751-4fab7a71b8ca
Baesens, B., Viaene, S., Vanthienen, J. and Dedene, G.
(2000)
Wrapped feature selection by means of guided neural network optimisation.
The Fifteenth International Conference on Pattern Recognition (ICPR'2000), Barcelona, Spain.
31 Aug 2000.
Record type:
Conference or Workshop Item
(Paper)
Abstract
In this paper, we discuss the implementation of a wrapped neural network feature selection approach, introduced here as the Weight Cascaded Retraining (WCR) algorithm. The first part of the paper provides an outline of the algorithm and elaborates on its formal underpinnings. Central to the whole feature pruning approach is the iteratively conceived guided function optimization realized by passing the optimized weight vector from one iteration step to the next. This essentially gives rise to a cascaded form of neural network retraining. In the second part of the paper the theoretical exposition of the WCR algorithm will be illuminated and benchmarked by means of publicly available UCI case material. It is illustrated that WCR based neural network feature selection may be very effective in reducing model complexity for classification modeling via neural networks.
This record has no associated files available for download.
More information
Published date: 2000
Venue - Dates:
The Fifteenth International Conference on Pattern Recognition (ICPR'2000), Barcelona, Spain, 2000-08-31 - 2000-08-31
Identifiers
Local EPrints ID: 36756
URI: http://eprints.soton.ac.uk/id/eprint/36756
PURE UUID: b3956e1c-380a-40e0-8f0a-2f0908ebf3ac
Catalogue record
Date deposited: 31 Jul 2006
Last modified: 09 Jan 2022 03:16
Export record
Contributors
Author:
S. Viaene
Author:
J. Vanthienen
Author:
G. Dedene
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics
