Abstract

Tolerance rough set model (TRSM) is based on rough set model with tolerance relation; due to the wide application based on tolerance relation, many researches on TRSM are increasing more. With the background of TRSM, the paper discusses how to determine the relevance features and noise data by the upper and lower approximations in the RELIEF algorithm, and the sequence of computing features according to the feature dependency. The proposed algorithm in the paper is given with reference data to validate the results of UCI after calculating the instance data, in enhancing the efficiency of the algorithm, and also effectively identifying important features and redundant features. The algorithm opens up a new application of feature selection with TRSM, which will be the coming research direction.

Paper Details