Title: Detecting malware in linguistic data using malware detection deep belief neural network method

Authors: M. Gomathy; A. Vidhya

Addresses: Department of Information Technology, Vels Institute of Science, Technology, and Advanced Studies, Pallavaram, Chennai, Tamil Nadu, India ' Department of Information Technology, Vels Institute of Science, Technology, and Advanced Studies, Pallavaram, Chennai, Tamil Nadu, India

Abstract: The widespread usage of high-end digital technologies has greatly increased cyber risks. To fight cybercrimes, a smart model should categorise and learn from data autonomously. Internet connectivity has made people's lifestyles more intertwined, and virtual collaboration is happening across regions. Pop-up messages also entice users and enable fraud. We use a neural network to predict unexpected pop-up message content in this paper. Modern malware and its powerful obfuscation algorithms have made traditional malware detection methods ineffective. However, deep belief neural networks (DBNNs) have garnered attention from researchers for malware detection to fight conventional cybercrime prevention methods in the long run. MDDBNN (malware detection deep belief neural network), based on file properties and contents, is proposed in this research for malware classification. The CLaMP Integrated dataset provided 5210 instances for training and testing. MDDBNN beats GaussianNB, LDA, logistic regression, and support vector machine (SVM). This study found that MDDBNN has the highest accuracy of 97.8%.

Keywords: deep belief networks; cyber security; cybercrime; spam and deep learning; DL; support vector machine; SVM; malware detection deep belief neural network; MDDBNN; logistic regression; LR.

DOI: 10.1504/IJCIS.2025.150792

International Journal of Critical Infrastructures, 2025 Vol.21 No.6, pp.640 - 662

Received: 30 Jan 2024
Accepted: 22 May 2024
Published online: 23 Dec 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article