NMGRS: Neighborhood-based multigranulation rough sets

doi:10.1016/j.ijar.2012.05.004

Return

NMGRS: Neighborhood-based multigranulation rough sets

2012-10-01

239

林国平 (Guoping Lin) *

钱宇华 (Yuhua Qian)

李锦锦 (Jinjin Li)

DOI：10.1016/j.ijar.2012.05.004

Original

Save

View PDF

Abstract

En 中文

Recently, a multigranulation rough set (MGRS) has become a new direction in rough set theory, which is based on multiple binary relations on the universe. However, it is worth noticing that the original MGRS can not be used to discover knowledge from information systems with various domains of attributes. In order to extend the theory of MGRS, the objective of this study is to develop a so-called neighborhood-based multigranulation rough set (NMGRS) in the framework of multigranulation rough sets. Furthermore, by using two different approximating strategies, i.e., seeking common reserving difference and seeking common rejecting difference, we first present optimistic and pessimistic 1-type neighborhood-based multigranulation rough sets and optimistic and pessimistic 2-type neighborhood-based multigranulation rough sets, respectively. Through analyzing several important properties of neighborhood-based multigranulation rough sets, we find that the new rough sets degenerate to the original MGRS when the size of neighborhood equals zero. To obtain covering reducts under neighborhood-based multigranulation rough sets, we then propose a new definition of covering reduct to describe the smallest attribute subset that preserves the consistency of the neighborhood decision system, which can be calculated by Chen's discernibility matrix approach. These results show that the proposed NMGRS largely extends the theory and application of classical MGRS in the context of multiple granulations. (C) 2012 Elsevier Inc. All rights reserved.

Keywords:

Rough sets

Granular computing

Attribute reduction

Multigranulation

Neighborhood relation

AI Summary

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