Local multigranulation decision-theoretic rough sets

Yuhua Qian

doi:10.1016/j.ijar.2016.12.008

Return

Local multigranulation decision-theoretic rough sets

2017-03-01

131

PRE

DOI：10.1016/j.ijar.2016.12.008

Original

Original request for help

Save

Abstract

En 中文

Multigranulation rough sets (MGRSs) where a target concept is approximated by granular structures induced by multiple binary relations have been applied successfully in many domains but they are still affected by two issues. First, similar to other rough set models, calculating the approximation of a target set is extremely time-consuming for larger scale data. Second, MGRSs comprise a supervised learning method, so they often require a large amount of labeled data. However, in the era of big data, labeling all data is almost infeasible in some cases. In this study, to address these issues, we propose the combination of local rough sets with multigranulation decision-theoretic rough sets to obtain local multigranulation decision-theoretic rough sets (CMG-DTRSs) as a semi unsupervised learning method. We also explore a number of important properties of LMG-DTRSs. In addition, we verify the efficiency of a concept approximation algorithm designed with. LMG-DTRS based on theoretical and experimental analyses. Furthermore, we present two types of local MGRS frameworks under PRSs and variable precision rough sets, where the relationships between them and the LMG-DTRS model are also discussed. We show that many local MGRS models can be derived from the LMG-DTRS framework. (C) 2016 Elsevier Inc. All rights reserved.

Keywords:

Decision-theoretic rough set

Limited labeled data

Local rough set

Multigranulation

Rough set

AI Summary

Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.