Enhancing cybersecurity: A machine learning-based system for malware attacks identify and classification
Multidisciplinary Science Journal · 09.10.2026 · 9 (4) · 2027396
DOI: 10.31893/multiscience.2027396 ↗Kurz gefasst
Fachartikel in der Zeitschrift Multidisciplinary Science Journal, online erschienen am 09.10.2026. Verfasst von Binu Nittadakkam und 5 weiteren Autor:innen. Die Arbeit nennt 45 Literaturangaben.
Abstract (Originalsprache)
The malware is one of the most pressing dangers of the modern digital ecosystem, aimed at the computers and other smart devices, as well as network infrastructures. With the rise in the use of internet, the malicious attacks using malware als o increases. The process of identifying malicious code is getting urgent, and the current methods should be enhanced. Conventional malware detection techniques may not scale to large-scale and multi-class datasets with high dimensionality and imbalance between classes hence decreasing accuracy and generalization. To overcome these drawbacks, the objective of this study is to design a powerful and scalable machine learning architecture to achieve accurate multi-class malware detection and classification. The proposed methodology will use the Total Features-ISCX Flow Meter dataset and apply the exploratory data analysis, label encoding, and feature reduction to improve the quality of data. The optimal choice feature selection is achieved with correlation analysis and SelectKBest (ANOVA F-scores), and SMOTE is used to address the issue of class imbalance. This is followed by comparing the enhanced dataset to more traditional algorithms such as SVM, BiGRU-CNN, Logistic Regression (LR), and Naive Bayes (NB), and then using them to train ensemble learning techniques like Random Forest (RF) and XGBoost (XGB).The perceptive quality and ROC curves as well as F1-scores, accuracy, and recall indicate the high discriminative power of the ensemble models especially RF. The results of the experiments prove that RF is the most accurate (94.15%), followed by XGB (93.48%), LR (92.85%), and NB (52.04%). The findings highlight the importance of the findings in their study, especially in the realization of cost-effective solutions to the detection of eight malware strains. The importance of the given study is that it offers a generalizable and high-performance malware detection framework that can potentially be used in managing real-world multi-class data with a high imbalance to improve cybersecurity defense mechanisms with a better model stability and accuracy. The suggested ML-based system provides a stable, understandable, and trustworthy multi-class malware detection model in a realistic cybersecurity setting.
Zitieren
APA
Nittadakkam, B., Abhishek, F., Shah, M. K., Pathak, H. K., Kumar, R., & Naranga, S. (2026). Enhancing cybersecurity: A machine learning-based system for malware attacks identify and classification. Multidisciplinary Science Journal, 9(4), 2027396. https://doi.org/10.31893/multiscience.2027396
Vancouver
Nittadakkam B, Abhishek F, Shah MK, Pathak HK, Kumar R, Naranga S. Enhancing cybersecurity: A machine learning-based system for malware attacks identify and classification. Multidisciplinary Science Journal. 2026;9(4):2027396. doi:10.31893/multiscience.2027396
BibTeX
@article{Nittadakkam2026,
author = {Nittadakkam, Binu and Abhishek, FNU and Shah, Maunik K. and Pathak, Harish Kant and Kumar, Rahul and Naranga, Surendra},
title = {Enhancing cybersecurity: A machine learning-based system for malware attacks identify and classification},
journal = {Multidisciplinary Science Journal},
year = {2026},
volume = {9},
number = {4},
pages = {2027396},
doi = {10.31893/multiscience.2027396}
}
EUVIMED