Bio-Inspired Feature Extraction and Ensemble Approach to Classify Malware

Malware classification plays an important role in preventing security threats. As malware affects many different areas like mobile, computer, IoT, etc., an effective classification approach needs to be used, even on big data samples. This paper gives a detailed analysis of ensemble architecture, highlighting the synergistic effect of combining CNNs with ACO and ANN for robust feature extraction and selection to provide a scalable solution for real-time malware detection. We propose a comprehensive model integrating a Convolutional Neural Network for feature extraction, enhanced with a Rectified Linear Unit (ReLU) combined with ACO for feature selection. In our experimental setup, to evaluate the effectiveness of the proposed model, we used the Microsoft BIG15 malware dataset with 9 different classes and obtained an accuracy of 98.76%, surpassing traditional and standalone methods.