A Correlation-Redundancy Guided Evolutionary Algorithm and Its Application to High-Dimensional Feature Selection in Classification

Abstract

The processing of high-dimensional datasets has become unavoidable with the development of information technology. Most of the literature on feature selection (FS) of high-dimensional datasets focuses on improvements in search strategies, ignoring the characteristics of the dataset itself such as the correlation and redundancy of each feature. This could degrade the algorithm's search effectiveness. Thus, this paper proposes a correlation-redundancy guided evolutionary algorithm (CRGEA) to address high-dimensional FS with the objectives of optimizing classification accuracy and the number of features simultaneously. A new correlation-redundancy assessment method is designed for selecting features with high relevance and low redundancy to speed up the entire evolutionary process. In CRGEA, a novel initialization strategy combined with a multiple threshold selection mechanism is developed to produce a high-quality initial population. A local acceleration evolution strategy based on a parallel simulated annealing algorithm and a pruning method is developed, which can search in different directions and perform deep searches combing the annealing stage around the best solutions to improve the local search ability. Finally, the comparison experiments on 16 public high-dimensional datasets verify that the designed CRGEA outperforms other state-of-the-art intelligent algorithms. The CRGEA can efficiently reduce redundant features while ensuring high accuracy.