Feature weighting as a tool for unsupervised feature selection

Deepak Panday; Renato Cordeiro De Amorim; Peter Lane

doi:10.1016/j.ipl.2017.09.005

Feature weighting as a tool for unsupervised feature selection

Deepak Panday, Renato Cordeiro De Amorim, Peter Lane

Research output: Contribution to journal › Article › peer-review

25 Citations (Scopus)

59 Downloads (Pure)

Abstract

Feature selection is a popular data pre-processing step. The aim is to remove some of the features in a data set with minimum information loss, leading to a number of benefits including faster running time and easier data visualisation.
In this paper we introduce two unsupervised feature selection algorithms. These make use of a cluster-dependent feature-weighting mechanism reflecting the within-cluster degree of relevance of a given feature. Those features with
a relatively low weight are removed from the data set. We compare our algorithms to two other popular alternatives using a number of experiments on both synthetic and real-world data sets, with and without added noisy features.
These experiments demonstrate our algorithms clearly outperform the alternatives.

Original language	English
Pages (from-to)	44-52
Number of pages	9
Journal	Information Processing Letters
Volume	129
Early online date	21 Sept 2017
DOIs	https://doi.org/10.1016/j.ipl.2017.09.005
Publication status	Published - 1 Jan 2018

Keywords

Algorithms
Clustering
Feature selection

Access to Document

10.1016/j.ipl.2017.09.005

Accepted ManuscriptAccepted author manuscript, 530 KBLicence: CC BY-NC-ND

Cite this

@article{755e5fae08234a3a9f375291a1af5ab0,

title = "Feature weighting as a tool for unsupervised feature selection",

abstract = "Feature selection is a popular data pre-processing step. The aim is to remove some of the features in a data set with minimum information loss, leading to a number of benefits including faster running time and easier data visualisation.In this paper we introduce two unsupervised feature selection algorithms. These make use of a cluster-dependent feature-weighting mechanism reflecting the within-cluster degree of relevance of a given feature. Those features witha relatively low weight are removed from the data set. We compare our algorithms to two other popular alternatives using a number of experiments on both synthetic and real-world data sets, with and without added noisy features.These experiments demonstrate our algorithms clearly outperform the alternatives.",

keywords = "Algorithms, Clustering, Feature selection",

author = "Deepak Panday and {Cordeiro De Amorim}, Renato and Peter Lane",

note = "This document is the Accepted Manuscript version of the following article: Deepak Panday, Renato Cordeiro de Amorin, and Peter Lane, {\textquoteleft}Feature weighting as a tool for unsupervised feature selection{\textquoteright}, Information Processing Letters, Vol. 129, January 2018. Under embargo. Embargo end date: 21 September 2018. Published by Elsevier.",

year = "2018",

month = jan,

day = "1",

doi = "10.1016/j.ipl.2017.09.005",

language = "English",

volume = "129",

pages = "44--52",

journal = "Information Processing Letters",

publisher = "Elsevier",

}

TY - JOUR

T1 - Feature weighting as a tool for unsupervised feature selection

AU - Panday, Deepak

AU - Cordeiro De Amorim, Renato

AU - Lane, Peter

N1 - This document is the Accepted Manuscript version of the following article: Deepak Panday, Renato Cordeiro de Amorin, and Peter Lane, ‘Feature weighting as a tool for unsupervised feature selection’, Information Processing Letters, Vol. 129, January 2018. Under embargo. Embargo end date: 21 September 2018. Published by Elsevier.

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Feature selection is a popular data pre-processing step. The aim is to remove some of the features in a data set with minimum information loss, leading to a number of benefits including faster running time and easier data visualisation.In this paper we introduce two unsupervised feature selection algorithms. These make use of a cluster-dependent feature-weighting mechanism reflecting the within-cluster degree of relevance of a given feature. Those features witha relatively low weight are removed from the data set. We compare our algorithms to two other popular alternatives using a number of experiments on both synthetic and real-world data sets, with and without added noisy features.These experiments demonstrate our algorithms clearly outperform the alternatives.

AB - Feature selection is a popular data pre-processing step. The aim is to remove some of the features in a data set with minimum information loss, leading to a number of benefits including faster running time and easier data visualisation.In this paper we introduce two unsupervised feature selection algorithms. These make use of a cluster-dependent feature-weighting mechanism reflecting the within-cluster degree of relevance of a given feature. Those features witha relatively low weight are removed from the data set. We compare our algorithms to two other popular alternatives using a number of experiments on both synthetic and real-world data sets, with and without added noisy features.These experiments demonstrate our algorithms clearly outperform the alternatives.

KW - Algorithms

KW - Clustering

KW - Feature selection

UR - http://www.scopus.com/inward/record.url?scp=85030177010&partnerID=8YFLogxK

U2 - 10.1016/j.ipl.2017.09.005

DO - 10.1016/j.ipl.2017.09.005

M3 - Article

VL - 129

SP - 44

EP - 52

JO - Information Processing Letters

JF - Information Processing Letters

ER -

Feature weighting as a tool for unsupervised feature selection

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this