Feature selection is an important processing step in machine learning and the design of pattern recognition system. Feature selection involves identifying a subset of the most useful features that produces compatible results as the original entire set of features. In this work, fuzzy entropy of each feature is calculated and filtered using linear search as well as redundancy is removed to judge on features suitability. The obtained results indicate that the generated features are of maximum relevance and with minimum degree of redundancy. Finally the proposed algorithm is applied to four different data sets, with four different threshold values and also compared against FCBF algorithm. Also the performance metrics (sensitivity, specificity and accuracy) and dimension metrics (features than the FCBF algorithm.selected) are compared for all the four data sets. All the results show that the proposed algorithm can give significant results for the feature selection process. Keywords: Fuzzy Entropy, Feature Selection, Symmetrical Uncertainty, Fuzzy Mutual Information, High Dimensional Data


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    An Efficient Fuzzy Entropy Based Feature Selection Algorithm for High Dimensional Data



    Erscheinungsdatum :

    2017-01-13


    Anmerkungen:

    doi:10.26483/ijarcs.v6i6.2534
    International Journal of Advanced Research in Computer Science; Vol 6, No 6 (2015): July - August 2015; 190-196 ; 0976-5697 ; 10.26483/ijarcs.v6i6



    Medientyp :

    Aufsatz (Zeitschrift)


    Format :

    Elektronische Ressource


    Sprache :

    Englisch


    Klassifikation :

    DDC:    006 / 629



    Feature selection for high-dimensional data in astronomy

    Zheng, Hongwen / Zhang, Yanxia | Elsevier | 2007


    Feature selection for high-dimensional data in astronomy

    Zheng, Hongwen | Online Contents | 2008


    Feature selection for high-dimensional data in astronomy

    Zheng, Hongwen | Online Contents | 2008



    Feature Selection for Classifying High-Dimensional Numerical Data

    Wu, Y. / Zhang, A. / IEEE Computer Society | British Library Conference Proceedings | 2004