Scholarlink

PDFs frequently contain malware, and the conventional methods of locating such threats do not always function for obfuscated files and zero-day attacks. This paper considers how machine learning (ML) can be utilized to locate malicious PDFs. We experimented with various ML techniques like MLP, KNN, SVM, Logistic Regression, and Random Forest using the Evasive-PDFMal2022 dataset. Our research reveals that identifying the right features and adjusting the hyperparameters is crucial in order to identify such threats adequately. The Random Forest algorithm with a limited number of features achieves a wonderful balance between accuracy and performance. These results indicate that machine learning is a powerful tool to assist in identifying risky PDF files.

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

Machine Learning-Based Detection of Malicious PDF Files: A Comparative Analysis

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Amal Shaji,
  • Ganesh Krishna,
  • C. B. Harigovind,
  • S. Gopika,
  • S. Remya,
  • Somula Ramasubbareddy

摘要

PDFs frequently contain malware, and the conventional methods of locating such threats do not always function for obfuscated files and zero-day attacks. This paper considers how machine learning (ML) can be utilized to locate malicious PDFs. We experimented with various ML techniques like MLP, KNN, SVM, Logistic Regression, and Random Forest using the Evasive-PDFMal2022 dataset. Our research reveals that identifying the right features and adjusting the hyperparameters is crucial in order to identify such threats adequately. The Random Forest algorithm with a limited number of features achieves a wonderful balance between accuracy and performance. These results indicate that machine learning is a powerful tool to assist in identifying risky PDF files.