Title: A graph-based approach for feature selection from higher order correlations

Authors: Sunanda Das; Asit Kumar Das

Addresses: Neotia Institute of Technology, Management and Science, Diamond Harbour, South 24-Pargana, 743368, WB, India ' Department of Computer Science and Technology, Indian Institute of Engineering Science and Technology, Shibpur, Howrah, 711103, WB, India

Abstract: Graph technology emerges as an important topic in the field of data mining and machine learning community. The analysis of high-dimensional data is crucial to identify a smaller subset of features which are informative for classification and clustering. In this paper, an efficient graph feature selection method is proposed to render the analysis of high-dimensional data tractable. Here, the feature scores are calculated for obtaining the weights of the edges in the weighted graph to identify the optimal feature subset. One advantage of this method is that it can successfully identify the optimal features for machine learning. The experimental results on our dataset verify the effectiveness and efficiency of the proposed method.

Keywords: graph technology; score; correlation; feature selection.

DOI: 10.1504/IJCSYSE.2018.090651

International Journal of Computational Systems Engineering, 2018 Vol.4 No.1, pp.66 - 71

Received: 19 Jan 2017
Accepted: 26 Aug 2017

Published online: 25 Mar 2018 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article