Different classes’ ratio fuzzy rough set based robust feature selection

In order to solve the problem that the classical fuzzy rough set (FRS) model used for feature selection is sensitive to noisy information, we propose an effective robust fuzzy rough set model, called different classes’ ratio fuzzy rough set (DC_ratio FRS) model. The proposed model can reduce the inf...

Full description

Saved in:
Bibliographic Details
Published in:Knowledge-based systems 2017-03, Vol.120, p.74-86
Main Authors: Li, Yuwen, Wu, Shunxiang, Lin, Yaojin, Liu, Jinghua
Format: Article
Language:English
Subjects:
Feature selection
Fuzzy rough set
Fuzzy sets
Noise
Robust
Robust control
Robustness (mathematics)
Rough set models
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In order to solve the problem that the classical fuzzy rough set (FRS) model used for feature selection is sensitive to noisy information, we propose an effective robust fuzzy rough set model, called different classes’ ratio fuzzy rough set (DC_ratio FRS) model. The proposed model can reduce the influence of noisy samples on the computation of the lower and upper approximations, and recognize the noisy samples directly. Moreover, the DC_ratio FRS model is robust against noise because it ignores a noisy sample which can be identified by computing the different classes’ ratio of this sample. Different classes’ ratio denotes the proportion of samples belonging to different classes in the neighbors of a given sample. Then, the properties of the DC_ratio FRS model are also discussed, and sample pair selection (SPS) based on the DC_ratio FRS model is used to feature selection. Finally, extensive experiments are given to illustrate the robustness and effectiveness of the proposed model.
ISSN:0950-7051
1872-7409
DOI:10.1016/j.knosys.2016.12.024