Volume 8 Issue 2 - March 2017

  • 1. Optimal feature selection algorithm for high dimensional data sets using particle swarm optimization

    Authors : D.sheela Jeyarani, Dr.mrs.a.pethalakshmi, Dr.mrs.k. Jayapriya

    Pages : 200-211

    DOI : http://dx.doi.org/10.21172/1.82.028

    Keywords : High Dimensional Data Set Fuzzy Entropy Particle Swam Optimization Feature Section

    Abstract :

    In high dimensional data sets, there are large numbers of features in classification problems. But all of them are not necessarily beneficial for classification. Feature selection involves the choosing of a small number of relevant features in order to achieve similar or even better classification performance. When all the features are used, irrelevant and redundant features reduce the performance. Maximizing the classification performance and minimizing the number of features are the important objectives of feature selection. Fuzzy entropy facilitates the partition of the input space into decision regions. It enables the selection of appropriate features with good seperability for the classification task. In the recent times, the observations in the swarm behavior of birds and fishes, has led to the study of the nature inspired heuristic algorithms called particle swarm optimization. It has been applied to find solution for many optimization applications. This paper proposes a hybrid method for feature subset selection based on Fuzzy entropy with particle swarm optimization. The proposed algorithm is applied to four different data sets. Further the performance metrics and the dimension metrics are calculated and compared with the existing FCBF (Fast correlation based filter solution) algorithm. The results show that the proposed algorithm can give significant results for the feature selection process.

    Citing this Journal Article :

    D.sheela Jeyarani, Dr.mrs.a.pethalakshmi, Dr.mrs.k. Jayapriya, "Optimal feature selection algorithm for high dimensional data sets using particle swarm optimization", Volume 8 Issue 2 - March 2017, 200-211