Intelligent AI Techniques for Obfuscated Privacy Malware Detection and Classification

View at publisher Summarise with VIRAT
Authors: Shakthi Sri P, Sobiya Devi G, M. A. Gunavathie
Cited by: 0
References: 15

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

The increasing dependence on digital systems has led to a rapid escalation in privacy-invasive malware, which now employs advanced obfuscation, encryption, and polymorphism to evade traditional signature-based defences. These evolving threats pose significant challenges to cybersecurity, as conventional detection mechanisms fail to identify sophisticated and previously unseen malware variants. To address this growing concern, intelligent, behaviour-driven detection methods have become essential. In this study, we propose a hybrid malware detection and classification framework designed to analyse memory-dump-based behavioural features from the CIC-MalMem-2022 dataset. The system incorporates a multi-level detection approach consisting of binary classification (benign vs. malicious), category-level classification (spyware, ransomware, trojan, benign), and family-level attribution across$\mathbf{1 6}$malware families. The architecture integrates machine learning models such as Random Forest with deep learning techniques like Deep Neural Networks (DNN) to capture complex behavioural patterns. Class imbalance is addressed using the Synthetic Minority Oversampling Technique (SMOTE), ensuring fair learning across minority malware categories and families. The proposed framework is deployed through a Flask-based web interface, enabling users to upload memory-dump features and receive real-time threat classification with confidence scores. Experimental results demonstrate high performance across all detection levels, with the hybrid ensemble model achieving up to 100 % accuracy in binary detection and 95.6 % accuracy in family-level classification, along with improved precision and recall, and strong robustness against obfuscated malware. These outcomes highlight the effectiveness of combining machine learning and deep neural networks for advanced malware detection.

Fields of Study & Subjects

Advanced Malware Detection Techniques Security and Verification in Computing Spam and Phishing Detection

Publication details

Journal / source
International Conference on Communication, Computing and Electronics Systems (ICCCES)
Pages
1447–1453
Publication date
21 January 2026
Language
EN