DEVELOPMENT OF A HYBRID NN–CNN DEEP LEARNING FRAMEWORK FOR INTELLIGENT MALWARE DETECTION, FAMILY CLASSIFICATION, AND VARIANT IDENTIFICATION
TL;DR
A Hybrid Neural Network–Convolutional Neural Network (NN–CNN) Deep Learning Framework for malware detection, malware-family classification, and malware-variant identification and considers two important issues in practical malware detection: model explainability and generalization to previously unseen malware.
Abstract
Malware continues to be one of the major cybersecurity threats affecting individuals, businesses, government institutions, and critical infrastructure. The problem has become more difficult because modern malware can change its structure, hide important characteristics, and produce multiple variants while maintaining similar malicious behaviour. Traditional signature-based detection methods remain useful for known threats, but they may struggle when presented with new or modified malware. This has encouraged researchers to explore artificial intelligence (AI) and deep learning as alternative approaches to automated malware analysis. This study proposes a Hybrid Neural Network–Convolutional Neural Network (NN–CNN) Deep Learning Framework for malware detection, malware-family classification, and malware-variant identification. The proposed framework combines two complementary learning approaches. The NN component will learn patterns from structured malware features, while the CNN component will extract spatial patterns from visual representations of malware binaries. The learned representations will then be combined through a feature-fusion mechanism and used for hierarchical classification. In the proposed approach, the system will first determine whether a file is benign or malicious, then identify the malware family, and finally attempt to determine the specific variant or subfamily. The proposed framework also considers two important issues in practical malware detection: model explainability and generalization to previously unseen malware. Explainable AI techniques will be investigated to help security analysts understand the factors influencing model decisions, while cross-dataset and temporal evaluations will be used to assess whether the model can maintain its performance beyond the dataset on which it was trained. The study is expected to contribute a more comprehensive approach to intelligent malware analysis by bringing detection, family classification, variant identification, explainability, and robustness into a single research framework.
