Data classification using the Dempster-Shafer method

In this paper, the Dempster-Shafer method is employed as the theoretical basis for creating data classification systems. Testing is carried out using three popular (multiple attribute) benchmark datasets that have two, three and four classes. In each case, a subset of the available data is used for...

Full description

Bibliographic Details
Main Authors: Chen, Qi, Whitbrook, Amanda, Aickelin, Uwe, Roadknight, Chris
Format: Article
Published: Taylor & Francis 2014
Subjects:
Online Access:https://eprints.nottingham.ac.uk/3346/
_version_ 1848791005928620032
author Chen, Qi
Whitbrook, Amanda
Aickelin, Uwe
Roadknight, Chris
author_facet Chen, Qi
Whitbrook, Amanda
Aickelin, Uwe
Roadknight, Chris
author_sort Chen, Qi
building Nottingham Research Data Repository
collection Online Access
description In this paper, the Dempster-Shafer method is employed as the theoretical basis for creating data classification systems. Testing is carried out using three popular (multiple attribute) benchmark datasets that have two, three and four classes. In each case, a subset of the available data is used for training to establish thresholds, limits or likelihoods of class membership for each attribute, and hence create mass functions that establish probability of class membership for each attribute of the test data. Classification of each data item is achieved by combination of these probabilities via Dempster’s Rule of Combination. Results for the first two datasets show extremely high classification accuracy that is competitive with other popular methods. The third dataset is non-numerical and difficult to classify, but good results can be achieved provided the system and mass functions are designed carefully and the right attributes are chosen for combination. In all cases the Dempster-Shafer method provides comparable performance to other more popular algorithms, but the overhead of generating accurate mass functions increases the complexity with the addition of new attributes. Overall, the results suggest that the D-S approach provides a suitable framework for the design of classification systems and that automating the mass function design and calculation would increase the viability of the algorithm for complex classification problems.
first_indexed 2025-11-14T18:21:38Z
format Article
id nottingham-3346
institution University of Nottingham Malaysia Campus
institution_category Local University
last_indexed 2025-11-14T18:21:38Z
publishDate 2014
publisher Taylor & Francis
recordtype eprints
repository_type Digital Repository
spelling nottingham-33462020-05-04T16:42:43Z https://eprints.nottingham.ac.uk/3346/ Data classification using the Dempster-Shafer method Chen, Qi Whitbrook, Amanda Aickelin, Uwe Roadknight, Chris In this paper, the Dempster-Shafer method is employed as the theoretical basis for creating data classification systems. Testing is carried out using three popular (multiple attribute) benchmark datasets that have two, three and four classes. In each case, a subset of the available data is used for training to establish thresholds, limits or likelihoods of class membership for each attribute, and hence create mass functions that establish probability of class membership for each attribute of the test data. Classification of each data item is achieved by combination of these probabilities via Dempster’s Rule of Combination. Results for the first two datasets show extremely high classification accuracy that is competitive with other popular methods. The third dataset is non-numerical and difficult to classify, but good results can be achieved provided the system and mass functions are designed carefully and the right attributes are chosen for combination. In all cases the Dempster-Shafer method provides comparable performance to other more popular algorithms, but the overhead of generating accurate mass functions increases the complexity with the addition of new attributes. Overall, the results suggest that the D-S approach provides a suitable framework for the design of classification systems and that automating the mass function design and calculation would increase the viability of the algorithm for complex classification problems. Taylor & Francis 2014-02-26 Article PeerReviewed Chen, Qi, Whitbrook, Amanda, Aickelin, Uwe and Roadknight, Chris (2014) Data classification using the Dempster-Shafer method. Journal of Experimental & Theoretical Artificial Intelligence . pp. 1-25. ISSN 0952-813X Data Mining Dempster–Shafer theory data classification Dempster's rule of combination http://dx.doi.org/10.1080/0952813X.2014.886301 doi:10.1080/0952813X.2014.886301 doi:10.1080/0952813X.2014.886301
spellingShingle Data
Mining
Dempster–Shafer theory
data classification
Dempster's rule of combination
Chen, Qi
Whitbrook, Amanda
Aickelin, Uwe
Roadknight, Chris
Data classification using the Dempster-Shafer method
title Data classification using the Dempster-Shafer method
title_full Data classification using the Dempster-Shafer method
title_fullStr Data classification using the Dempster-Shafer method
title_full_unstemmed Data classification using the Dempster-Shafer method
title_short Data classification using the Dempster-Shafer method
title_sort data classification using the dempster-shafer method
topic Data
Mining
Dempster–Shafer theory
data classification
Dempster's rule of combination
url https://eprints.nottingham.ac.uk/3346/
https://eprints.nottingham.ac.uk/3346/
https://eprints.nottingham.ac.uk/3346/