Affiliation:
1. Capitol Technology University, USA & Illinois Institute of Technology, USA
Abstract
Cybercrime has grown into a multi-billion dollar industry in recent years. Malware deployment is one of these cybercrimes' most common aspects. This malicious software has shown its ability to cripple large commercial organizations and collect significant financial tolls up to billions of dollars yearly. It targets a variety of industries, sectors, corporations, and even individual entities without discrimination. Malware writers continuously hone their techniques and raise the bar on their sophistication, creating difficult-to-detect malware that may be left unnoticed in the background for extended periods of time to get around security measures. The first accuracy rate of the baseline model is a phenomenal 98%. The accuracy of the CNN model increases to an astonishing 99.183% by increasing its complexity, outperforming the performance of the bulk of CNN models reported in the literature. This CNN model is used to forecast the appearance of new malware samples in our dataset, further demonstrating its effectiveness.
Reference51 articles.
1. Obfuscated Malware Detection in IoT Android Applications Using Markov Images and CNN
2. Behavior-based ransomware classification: A particle swarm optimization wrapper-based approach for feature selection
3. AgarwalB. (2020). Deep Learning Techniques for Biomedical and Health Informatics. Academic Press.
4. Ahmadi, D. (2016). Novel Feature Extraction, Selection and Fusion for Effective Malware Family Classification. Proc. of the 6th ACM Conf. on Data and Application Sec. and Privacy.
5. Alazab, M., Layton, R., Venkatraman, S., & Watters, P. (2020). Malware detection based on structural and behavioral features. In Seventh Australasian Workshop on Software and System Architectures (Vol. 13, No. 2, pp. 1-11). Academic Press.