Download PDFOpen PDF in browserA Comprehensive Review on Machine Learning and Deep Learning Based Malware Detection MethodsEasyChair Preprint 159908 pages•Date: August 7, 2025AbstractMalware detection has become a significant aspect of cybersecurity, specifically with the widespread use of Android devices. Conventional malware detection methods, such as static and signature-based approaches have been overtaken by evolving and increasingly sophisticated malware techniques. Machine learning and deep learning have emerged as powerful tools in this field, offering enhanced detection capabilities through advanced pattern recognition. This paper presents a comprehensive review of various machine learning and deep learning methods, including convolutional neural networks (CNN), Bayesian classification, ensemble learning, and hybrid models for malware detection. The study evaluates these techniques in terms of accuracy, efficiency, adaptability, and their capacity to handle real-time detection, dataset diversity, and obfuscated malware. Additionally, it explores challenges such as class imbalance and the need for more interpretable models. The findings suggest that while CNN-based methods offer the highest accuracy, ensemble models strike a balance between precision and computational efficiency. Keyphrases: Android Malware Detection, Mobile Security, deep learning, hybrid models, machine learning, real-time detection
|