Scholarlink

This research proposes a hybrid multi-objective feature selection framework using NSGA-II combined with the information-theoretic filter method DRJMIM. The approach balances conflicting goals, minimizing selected features and maximizing classification accuracy by integrating attribute multicorrelation, complementarity, and redundancy. To enhance population diversity and convergence, a sinusoidal chaotic map is embedded into NSGA-II, and population initialization is guided by DRJMIM-ranked features. The method is applied to 5 different datasets including cervical cancer risk factor prediction, a relatively underexplored domain. Performance is evaluated using Hypervolume (HV) and Hamming distance, showing that the chaotic NSGA-II outperforms the standard NSGA-II in convergence and diversity.

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

IN2DFS: A Hybrid Multiobjective Feature Selection Model with Chaotic NSGA-II and DRJMIM Filter Method

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Sangeeta Kurman,
  • Sumitra Kisan

摘要

This research proposes a hybrid multi-objective feature selection framework using NSGA-II combined with the information-theoretic filter method DRJMIM. The approach balances conflicting goals, minimizing selected features and maximizing classification accuracy by integrating attribute multicorrelation, complementarity, and redundancy. To enhance population diversity and convergence, a sinusoidal chaotic map is embedded into NSGA-II, and population initialization is guided by DRJMIM-ranked features. The method is applied to 5 different datasets including cervical cancer risk factor prediction, a relatively underexplored domain. Performance is evaluated using Hypervolume (HV) and Hamming distance, showing that the chaotic NSGA-II outperforms the standard NSGA-II in convergence and diversity.