Instance reduction for one-class classification

Instance reduction techniques are data preprocessing methods originally developed to enhance the nearest neighbor rule for standard classification. They reduce the training data by selecting or generating representative examples of a given problem. These algorithms have been designed and widely anal...

Full description

Bibliographic Details
Main Authors: Krawczyk, Bartosz, Triguero, Isaac, García, Salvador, Woźniak, Michał, Herrera, Francisco
Format: Article
Published: Springer 2018
Subjects:
Online Access:https://eprints.nottingham.ac.uk/52077/
_version_ 1848798641880301568
author Krawczyk, Bartosz
Triguero, Isaac
García, Salvador
Woźniak, Michał
Herrera, Francisco
author_facet Krawczyk, Bartosz
Triguero, Isaac
García, Salvador
Woźniak, Michał
Herrera, Francisco
author_sort Krawczyk, Bartosz
building Nottingham Research Data Repository
collection Online Access
description Instance reduction techniques are data preprocessing methods originally developed to enhance the nearest neighbor rule for standard classification. They reduce the training data by selecting or generating representative examples of a given problem. These algorithms have been designed and widely analyzed in multi-class problems providing very competitive results. However, this issue was rarely addressed in the context of one-class classification. In this specific domain a reduction of the training set may not only decrease the classification time and classifier’s complexity, but also allows us to handle internal noisy data and simplify the data description boundary. We propose two methods for achieving this goal. The first one is a flexible framework that adjusts any instance reduction method to one-class scenario by introduction of meaningful artificial outliers. The second one is a novel modification of evolutionary instance reduction technique that is based on differential evolution and uses consistency measure for model evaluation in filter or wrapper modes. It is a powerful native one-class solution that does not require an access to counterexamples. Both of the proposed algorithms can be applied to any type of one-class classifier. On the basis of extensive computational experiments, we show that the proposed methods are highly efficient techniques to reduce the complexity and improve the classification performance in one-class scenarios.
first_indexed 2025-11-14T20:23:00Z
format Article
id nottingham-52077
institution University of Nottingham Malaysia Campus
institution_category Local University
last_indexed 2025-11-14T20:23:00Z
publishDate 2018
publisher Springer
recordtype eprints
repository_type Digital Repository
spelling nottingham-520772020-05-04T19:37:13Z https://eprints.nottingham.ac.uk/52077/ Instance reduction for one-class classification Krawczyk, Bartosz Triguero, Isaac García, Salvador Woźniak, Michał Herrera, Francisco Instance reduction techniques are data preprocessing methods originally developed to enhance the nearest neighbor rule for standard classification. They reduce the training data by selecting or generating representative examples of a given problem. These algorithms have been designed and widely analyzed in multi-class problems providing very competitive results. However, this issue was rarely addressed in the context of one-class classification. In this specific domain a reduction of the training set may not only decrease the classification time and classifier’s complexity, but also allows us to handle internal noisy data and simplify the data description boundary. We propose two methods for achieving this goal. The first one is a flexible framework that adjusts any instance reduction method to one-class scenario by introduction of meaningful artificial outliers. The second one is a novel modification of evolutionary instance reduction technique that is based on differential evolution and uses consistency measure for model evaluation in filter or wrapper modes. It is a powerful native one-class solution that does not require an access to counterexamples. Both of the proposed algorithms can be applied to any type of one-class classifier. On the basis of extensive computational experiments, we show that the proposed methods are highly efficient techniques to reduce the complexity and improve the classification performance in one-class scenarios. Springer 2018-05-21 Article PeerReviewed Krawczyk, Bartosz, Triguero, Isaac, García, Salvador, Woźniak, Michał and Herrera, Francisco (2018) Instance reduction for one-class classification. Knowledge and Information Systems . ISSN 0219-1377 Machine learning ; One-class classification ; Instance reduction ; Training set selection ; Evolutionary computing https://link.springer.com/article/10.1007%2Fs10115-018-1220-z doi:10.1007/s10115-018-1220-z doi:10.1007/s10115-018-1220-z
spellingShingle Machine learning ; One-class classification ; Instance reduction ; Training set selection ; Evolutionary computing
Krawczyk, Bartosz
Triguero, Isaac
García, Salvador
Woźniak, Michał
Herrera, Francisco
Instance reduction for one-class classification
title Instance reduction for one-class classification
title_full Instance reduction for one-class classification
title_fullStr Instance reduction for one-class classification
title_full_unstemmed Instance reduction for one-class classification
title_short Instance reduction for one-class classification
title_sort instance reduction for one-class classification
topic Machine learning ; One-class classification ; Instance reduction ; Training set selection ; Evolutionary computing
url https://eprints.nottingham.ac.uk/52077/
https://eprints.nottingham.ac.uk/52077/
https://eprints.nottingham.ac.uk/52077/