A Fine-Grained Random Forests using Class Decomposition

Elyan, Eyad and Gaber, Mohamed Medhat (2016) A Fine-Grained Random Forests using Class Decomposition. Neural Computing and Applications, 27 (8). p. 2279. ISSN 0941-0643

3829.pdf - Accepted Version

Download (227kB)


Class decomposition describes the process of segmenting each class into a number of homogeneous subclasses. This can be naturally achieved through clustering. Utilising class decomposition can provide a number of benefits to supervised learning, especially ensembles. It can be a computationally efficient way to provide a linearly separable dataset without the need for feature engineering required by techniques like Support Ve]ctor Machines
(SVM) and Deep Learning. For ensembles, the decomposition is a natural way to increase diversity; a key factor for the success of ensemble classifiers. In this
paper, we propose to adopt class decomposition to the state-of-the-art ensemble learning Random Forests. Medical data for patient diagnosis may greatly benefit from this technique, as the same disease can have a diverse of symptoms. We have experimentally validated our proposed method on a number of datasets in that are mainly related to the medical domain. Results reported
in this paper shows clearly that our method has significantly improved the accuracy of Random Forests.

Item Type: Article
Additional Information: The final publication is available at link.springer.com via https://doi.org/10.1007/s00521-015-2064-z
Identification Number: https://doi.org/10.1007/s00521-015-2064-z
22 September 2015Published Online
November 2016Published
8 September 2015Accepted
Uncontrolled Keywords: Machine learning · Random Forests · Clustering · Ensemble learning
Subjects: CAH11 - computing > CAH11-01 - computing > CAH11-01-01 - computer science
Divisions: Faculty of Computing, Engineering and the Built Environment
Faculty of Computing, Engineering and the Built Environment > School of Computing and Digital Technology
Depositing User: Ian Mcdonald
Date Deposited: 26 Jan 2017 11:44
Last Modified: 22 Mar 2023 12:01
URI: https://www.open-access.bcu.ac.uk/id/eprint/3829

Actions (login required)

View Item View Item


In this section...