.png){kind=icon}
.png){kind=icon}
Return
ODS-EA: An objective to decision space-based evolutionary algorithm for high-dimensional feature selection
.png){kind=icon}
PRE
AI
DOI：10.1016/j.eswa.2025.130560.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Evolutionary algorithms (EAs) have shown their competitiveness in solving feature selection problems. However, due to the “curse of dimensionality”, it is quite challenging for EAs to tackle the high-dimensional feature selection (HDFS). To this end, recently, several efforts have been made to design EAs for HDFS. Unlike existing works that solve the challenge from the decision space, in this paper, we tackle the issue from a new perspective: objective space. Specifically, an objective to decision space-based evolutionary algorithm (ODS-EA) is proposed for HDFS, where the feature subsets are obtained by two-stage search in the objective space and the decision space, respectively. In the first stage of ODS-EA, a (1 × 1) objective space is first constructed for feature selection. Then, a subregion-based evolutionary strategy is suggested to search in the objective space, which can solve the “curse of dimensionality” effectively and select the latent good feature combinations. In the second stage, the search is performed on the decision space, where a feature contribution-based improving strategy is developed to enhance the solutions in the first stage and achieve the final feature subsets with higher quality. Experimental results on 12 high-dimensional datasets demonstrate the effectiveness and efficiency of the proposed ODS-EA, when compared with the state-of-the-art.
Journal
IF:
7.5
Papers:
2.9W
Citations:
10.2W