ULBP-RF: A Hybrid Approach for Malware Image Classification
Shakshi Gupta, Priti Bansal, Sumit Kumar · 2018
The growing dependence on internet for performing critical activities in every domain has raised serious concerns about the security of the computer systems. Malwares have become a significant threat to computer systems and recently, a massive growth has been observed by experts in the number and sophistication of new malwares. Therefore, the task of malware detection and classification is of utmost importance. However, the task of classifying malwares has become more challenging since the introduction of code obfuscation and metamorphism techniques. These techniques easily alter the malwares' code signatures and make the static techniques of malware detection ineffective. Dynamic analysis is effective but time-consuming. Recently, image processing techniques along with machine learning techniques have been explored by researchers for visualization and classification of malwares. In this paper, we introduce a new approach ULBP-RF that uses Uniform local binary pattern with circular neighborhood strategy to extract features from the malware images dataset. The resulting datasets are classified using Random Forest. To assess the performance of the proposed approach, experiments are performed and a comparative analysis of the performance of various combinations of feature extraction techniques and classification algorithms is done. It has been found that ULBP-RF has the highest classification accuracy.