A Parallel Approach for RegularSearch Algorithm

The process of finding minimal subsets of features that can differentiate objects belonging to different classes, known as typical testors, is an exponential complexity problem. Several algorithms have been proposed to improve the search process efficiency by utilizing different properties and techniques, including parallelism. In this paper, we propose a parallel version of the RegularSearch algorithm to find all typical testors related to a supervised classification problem. The proposed algorithm is compared with the most recent algorithm in the literature, and the comparative analysis shows the advantages of our proposal over the compared methods, using both synthetic and real problem datasets.