Deep Learning Based Maldroid Stacked Propagate Network for Android Malware Prediction for Security Enhancement

Abstract

Listen

Translate article

Objectives: This study aims to address the growing challenge of Android malware obfuscation, which undermines traditional detection methods. We propose the MalDroid Stacked Propagate Network (MDSPN), a deep learning-based system designed to enhance the accuracy and effectiveness of Android malware detection, with a focus on overcoming obfuscation techniques and improving classification performance. Methods: The MDSPN framework incorporates advanced data preprocessing techniques, including regressor data imputation to handle missing values and Adam Divisive Clustering to effectively organize and categorize malware features. Findings: MDSPN achieved an outstanding 99.5% detection accuracy, surpassing traditional state-of-the-art malware detection techniques by 7.3% in classification accuracy. Additionally, the model demonstrated a 92% precision rate and reduced false positives by 15%, highlighting its superior performance in identifying and classifying malicious Android applications. Novelty: This study introduces an innovative approach by combining data imputation, divisive clustering, and deep learning within the MDSPN framework. The integration of these techniques provides a more accurate, computationally efficient alternative to conventional detection methods, advancing the state-of-the-art in Android malware detection and offering a robust solution to combat sophisticated malware obfuscation strategies. Keywords: Android Malware Detection (AMD), MalDroid stacked propagate network, Androzoo and Drebin Dataset, Malware Classification, Deep Learning