Hybrid CNN-GRU Architecture for Malware Detection Using Memory Forensic Datasets

Abstract

Malware detection is a critical aspect requiring robust and adaptive techniques to identify evolving threats. The rapid proliferation of IoT devices has led to exponential growth. Machine learning methods, though effective, often struggle to capture the complex spatial. This study proposed a hybrid deep learning model that integrates CNN with Gated Recurrent Units (GRU) to enhance the classification performance on IoT datasets. The architecture leverages the spatial feature extraction capabilities of CNNs and the temporal sequence modeling strength of GRUs. The preprocessing pipeline includes data cleaning, feature extraction, and feature selection, followed by a data-splitting strategy to ensure robust training and evaluation. The proposed model achieves superior classification accuracy Furthermore, performance evaluation using training/testing accuracy and loss metrics demonstrates the model's stability, efficiency, and generalization ability. The findings suggest that the proposed method offers a highly reliable and scalable solution for IoT data analysis.