Detecting Malware in PDF Files: A Comparative Analysis of Machine Learning and Deep Learning Approaches

Machine learning (ML) has revolutionized malware detection, offering advanced methods to identify and mitigate threats with high precision. In the domain of PDF malware detection, ML models have demonstrated exceptional effectiveness in detecting subtle indicators of malicious content, thereby enhancing the security of digital documents. This study presents a structured approach to distinguishing between benign and malicious PDF files using various ML algorithms and deep neural networks (DNN). The approach encompassed thorough data preprocessing followed by the implementation of various machine learning models, such as Random Forest, Deep Neural Networks (DNN), Decision Tree, and K-Nearest Neighbors (KNN), among others. The models' performance was assessed using common evaluation metrics, including precision, recall, F1-score, and accuracy. The results highlight that the Random Forest classifier achieved the highest accuracy at 99.74%, outperforming other models. The DNN also performed impressively, closely following Random Forest in accuracy. Decision Tree and KNN demonstrated high accuracy as well, further validating their effectiveness in PDF malware detection. The results highlight the effectiveness of ensemble methods and advanced neural networks in precisely detecting and classifying malicious PDF files, thereby supporting the creation of strong cybersecurity defenses against PDF-based threats.