Rough sets attribute reduction using an accelerated genetic algorithm

Abstract

Attribute reduction is the process of removing a subset of attributes from the dataset. One of the most famous tools used for solving the attribute reduction problem is rough set theory. The current attribute reduction methods in rough set theory are failed for finding the optimal reduction because of no perfect heuristic can ensure optimality. In this paper, we consider a novel rough set approach to attribute reduction based on heuristic genetic algorithm. The proposed method, called accelerated genetic algorithm attribute reduction (AGAAR). The proposed method uses new suitable crossover and mutation operators that fit the considered problem. Moreover, an acceleration technique is also invoked in order to accelerate the search process for the optimal reduction. The experiment is archived to AGAAR through 13 well-known datasets from UCI machine learning repository. The experiment proves that the algorithm is more effective, it has improved the global search ability to avoid falling into local optimum, and it can get relative minimum attribute reduction.