Scholarlink

<p>In machine learning, the importance of relevant data increases exponentially. In our proposed approach, we introduce an optimization method that combines Particle Swarm Optimization (PSO) and the Firefly Algorithm (FA) to enhance feature selection using decision tree-based classification. PSO is well-suited for small search spaces, while the Firefly Algorithm is effective for large search spaces. The proposed method, PSOFA-DT aims to improve classification performance by reducing dimensionality and optimizing feature selection. PSO’s global search capabilities are complemented by FA’s localized search, and the algorithm’s effectiveness is evaluated using decision tree accuracy and hold-out cross-validation. Experimental results demonstrate that PSOFA-DT outperforms individual implementations of PSO and FA in feature reduction and classification accuracy. Decision tree accuracy is used as the primary fitness metric, while the Firefly Algorithm refines the feature selection process. The algorithm balances exploration and exploitation by adjusting key parameters such as inertia weight, learning factors, and attraction coefficients. The Firefly Algorithm further optimizes feature selection, enhancing decision tree performance.</p>

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

Hybrid multi-objective particle swarm optimization feature selection approach with firefly algorithm using decision tree classifier

  • 期刊论文

  • 发表日期：

    2025年3月21日

查看全文
  • Ashish Kumar Singh,
  • Anoj Kumar

摘要

In machine learning, the importance of relevant data increases exponentially. In our proposed approach, we introduce an optimization method that combines Particle Swarm Optimization (PSO) and the Firefly Algorithm (FA) to enhance feature selection using decision tree-based classification. PSO is well-suited for small search spaces, while the Firefly Algorithm is effective for large search spaces. The proposed method, PSOFA-DT aims to improve classification performance by reducing dimensionality and optimizing feature selection. PSO’s global search capabilities are complemented by FA’s localized search, and the algorithm’s effectiveness is evaluated using decision tree accuracy and hold-out cross-validation. Experimental results demonstrate that PSOFA-DT outperforms individual implementations of PSO and FA in feature reduction and classification accuracy. Decision tree accuracy is used as the primary fitness metric, while the Firefly Algorithm refines the feature selection process. The algorithm balances exploration and exploitation by adjusting key parameters such as inertia weight, learning factors, and attraction coefficients. The Firefly Algorithm further optimizes feature selection, enhancing decision tree performance.