Abstract

Malware detection has been an ongoing challenge for cybersecurity experts due to the evolving nature of malicious software and the ability of malware to disguise itself. Traditional methods that rely solely on static features such as file signatures or dynamic analysis have had limitations in detecting new or obfuscated malware. This paper investigates the enhancement of malware detection by integrating both static and dynamic features and utilizing deep neural networks (DNNs) for more effective classification. By combining these feature sets, this method aims to overcome the limitations of each individual feature type and improve the detection rate, minimize false positives, and increase the robustness of malware detection systems. The study also explores the challenges of dataset construction, preprocessing techniques, and the performance evaluation of DNN-based models for malware detection.