Scholarlink

This paper aims to propose an improved method for enhancing malware detection in portable executables (PE), through the Random Forest and variance-based feature selection mechanisms, in order to achieve higher accuracy than previous models, while decreasing training time. Our method involves training a Random Forest model with default parameters, its evaluation on the EMBER2018 portable executables dataset, and selection of relevant columns with variance exceeding a proposed threshold. We present our method’s results, compare them with those of previous works done in the literature, and outline our approach’s empirical benefits and reduced training time. Finally, we discuss potential ideas for future work and further improvements.

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

Enhancing Malware Detection In Portable Executables with Random Forest and Variance-Based Feature Selection

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Alexandru Todea,
  • Ciprian Pungilă,
  • Adrian Spătaru

摘要

This paper aims to propose an improved method for enhancing malware detection in portable executables (PE), through the Random Forest and variance-based feature selection mechanisms, in order to achieve higher accuracy than previous models, while decreasing training time. Our method involves training a Random Forest model with default parameters, its evaluation on the EMBER2018 portable executables dataset, and selection of relevant columns with variance exceeding a proposed threshold. We present our method’s results, compare them with those of previous works done in the literature, and outline our approach’s empirical benefits and reduced training time. Finally, we discuss potential ideas for future work and further improvements.