Scholarlink

The most vital topic in the machine learning field is feature selection (FS). The classifier’s ability is directly related to the excellence of feature selection. The feature subset search process, the feature subset evaluation process, the terminating criteria and the authentication method are the parts of the feature selection process. In this study, Pareto-based forest optimization technique is applied for the FS process by integrating wrapper and filter approaches (PFOA_WF). The aim of this study is to reduce the feature count and identify the related information to develop the classification accuracy. A local search (LS) selection strategy is applied to classify the population with Pareto front solutions, which enhances the solution set. The proposed PFOA_WF approach simultaneously optimizes the fitness function. The capabilities of PFOA_WF are validated using UCI repository datasets. The outcomes of PFOA_WF are compared with some multi-objective approaches. The results show that the PFOA_WF is more optimal than the prevailing optimization approaches.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Pareto Optimality Technique for Feature Selection by Integrating Wrapper-Filter Approach

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • G. Anitha,
  • B. Rosiline Jeetha

摘要

The most vital topic in the machine learning field is feature selection (FS). The classifier’s ability is directly related to the excellence of feature selection. The feature subset search process, the feature subset evaluation process, the terminating criteria and the authentication method are the parts of the feature selection process. In this study, Pareto-based forest optimization technique is applied for the FS process by integrating wrapper and filter approaches (PFOA_WF). The aim of this study is to reduce the feature count and identify the related information to develop the classification accuracy. A local search (LS) selection strategy is applied to classify the population with Pareto front solutions, which enhances the solution set. The proposed PFOA_WF approach simultaneously optimizes the fitness function. The capabilities of PFOA_WF are validated using UCI repository datasets. The outcomes of PFOA_WF are compared with some multi-objective approaches. The results show that the PFOA_WF is more optimal than the prevailing optimization approaches.