A new soft set based pruning algorithm for ensemble method
Ensemble methods have been introduced as a useful and effective solution to improve the performance of the classification. Despite having the ability of producing the highest classification accuracy, ensemble methods have suffered significantly from their large volume of base classifiers. Neverthele...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Asian Research Publishing Network
2016
|
| Subjects: | |
| Online Access: | http://eprints.unisza.edu.my/7489/1/FH02-FIK-16-06166.jpg http://eprints.unisza.edu.my/7489/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Universiti Sultan Zainal Abidin |
| Language: | English |
| Summary: | Ensemble methods have been introduced as a useful and effective solution to improve the performance of the classification. Despite having the ability of producing the highest classification accuracy, ensemble methods have suffered significantly from their large volume of base classifiers. Nevertheless, we could overcome this problem by pruning some of the classifiers in the ensemble repository. However, only a few researches focused on the ensemble pruning algorithm. Therefore, this paper aims to increase classification accuracy and at the same time minimizing ensemble classifiers by constructing a new ensemble pruning method (SSPM) based on dimensionality reduction in soft set theory. Ensemble pruning deals with the reduction of predictive models in order to improve its efficiency and predictive performance. Soft set theory has been proved to be an effective mathematical tool for dimension reduction. Thus, we proposed a novel soft set based method to prune the classifiers from heterogeneous ensemble committee and select the best subsets of the component classifiers prior to the combination process. The results show that the proposed method not only reduce the number of members of the ensemble, but able to produce highest prediction accuracy. |
|---|