CCFS: A cooperating coevolution technique for large scale feature selection on microarray datasets. (April 2018)

• Record Type: Journal Article
• Authors: Ebrahimpour, Mohammad K.; Nezamabadi-pour, Hossein; Eftekhari, Mahdi
• Is Part Of: Computational biology and chemistry. Volume 73(2018)
• Page Start: 171
• Page End: 178
• Publication Date: 2018-04
• Subjects: Meta-heuristics -- Cooperating coevolving feature selection -- Microarray datasets -- Divide and conquered algorithms; Chemistry -- Data processing -- Periodicals; Biology -- Data processing -- Periodicals; Biochemistry -- Data processing; Biology -- Data processing; Molecular biology -- Data processing; Periodicals; Electronic journals; 542.85
• DOI: 10.1016/j.compbiolchem.2018.02.006
• Languages: English
• ISSNs: 1476-9271

Abstract

Graphical abstract: Highlights: Proposing a Cooperative Coevolving (CC) version of BGSA. Using the CC version of BGSA in order to propose a novel global search for feature selection in high dimensional datasets. Proposing a novel divide and conquer feature selection algorithm based on CC. A comprehensive comparison with nine state-of-the-art algorithms on seven microarray high dimensional datasets is done and results are presented. Abstract: Recently, advances in bioinformatics lead to microarray high dimensional datasets. These kinds of datasets are still challenging for researchers in the area of machine learning since they suffer from small sample size and extremely large number of features. Therefore, feature selection is the problem of interest in the learning process in this area. In this paper, a novel feature selection method based on a global search (by using the main concepts of divide and conquer technique) which is called CCFS, is proposed. The proposed CCFS algorithm divides vertically (on features) the dataset by random manner and utilizes the fundamental concepts of cooperation coevolution by using a filter criterion in the fitness function in order to search the solution space via binary gravitational search algorithm. For determining the effectiveness of the proposed method some experiments are carried out on seven binary microarray high dimensional datasets. The obtained results are compared with nine state-of-the-art feature selection algorithms including Interact (INT), and Maximum Relevancy Minimum Redundancy (MRMR). The average outcomes of the results are analyzed by a statistical non-parametric test and it reveals that the proposed method has a meaningful difference to the others in terms of accuracy, sensitivity, specificity and number of selected features.