Survey of Deep Learning Techniques for Malware Detection: Insights, Challenges, and Future Directions

Android and Windows become the main platforms for mobile and personal computing, the threat from advanced malware targeting these systems is increasing. Traditional machine learning methods are struggling to keep up with these sophisticated threats, so deep learning techniques, which have shown better results, are needed. This paper reviews various deep learning methods for detecting and classifying malware across different environments, like IoT and web platforms. We examined 40 research articles from major digital libraries, focusing on different deep learning models like CNNs, LSTMs, DBNs, and autoencoders.Our analysis highlights the strengths and weaknesses of these models, such as high false positive rates and difficulties with detecting hidden malware. In order to decrease false positives and increase detection accuracy, we provide a novel framework that integrates many deep learning methods. This survey offers a comprehensive guide for new researchers and practical insights for improving malware detection with deep learning.