Static and Dynamic Malware Analysis Using Machine Learning

Abstract

The necessity for efficient, automated detection systems that can accurately identify malicious software has been brought to light by the growing sophistication of malware assaults. Despite their value, conventional static as well as dynamic evaluation tools sometimes lack the flexibility to adjust to changing infection strategies. In order to improve detection capabilities, this study employs a machine learning-based strategy that combines static and dynamic analysis of malware. To guarantee data quantity and importance for analysis, the system uses a large malware dataset and then applies data preparation techniques including feature scaling and normalization. The Extra-Trees-Classifier streamlines the classification process by identifying the most informative features through feature selection optimization. A Random Forest model, the main classifier, is used to evaluate the generated data and categorize files as either malware-free or infected. With a high precision of 99.42%, this model effectively and with little mistake distinguishes between harmful and benign files. This system offers a dependable, high-performance solution for proactive malware identification, which is crucial for contemporary cybersecurity applications, by fusing strong feature engineering with sophisticated classification approaches.