Anzeige
Erschienen in:
2018 | OriginalPaper | Buchkapitel
A Novel Attribute Reduction Approach Based on Improved Attribute Significance
verfasst von : Jun Ye, Lei Wang
Erschienen in: Computational Intelligence and Intelligent Systems
Verlag: Springer Singapore
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
Abstract
Aiming at the limitations of attribute reduction approach based on Pawlak attribute significance and conditional entropy, an efficient attribute reduction algorithm based on distinguish matrix and the improved attribute significance is put forward. Firstly, the deficiencies of two kinds of classical attribute reduction are analyzed. Furthermore, an improved attribute significance definition is given according to ability to distinguish one object from another in the universe, and then it is used to calculate the significance of attribute in discernibility matrix; Finally, a minimum attribute reduction can be gained by adding attribute to the core attribute set one by one according to descending order of attribute significance. Analysis on numerical example shows that the proposed algorithm can find the minimal attribute reduction effectively. Compared with the previous algorithm, the proposed algorithm can reduce the calculation amount on reduction greatly in the decision table which has more conditional attributes.