AMDDLmodel: Android smartphones malware detection using deep learning model.

Area of Science:

• Computer Science
• Cybersecurity
• Artificial Intelligence

Background:

• Android's widespread use makes its ecosystem a target for malware.
• Malware installation occurs through various vectors like API calls and permission grants, compromising user privacy and system security.
• Existing methods for Android malware detection and classification require enhancement to combat sophisticated threats.

Purpose of the Study:

• To develop and evaluate AMDDLmodel, a novel deep learning technique for accurate Android malware detection and classification.
• To enhance the security of Android devices and protect user privacy from malicious applications.
• To demonstrate the effectiveness of deep learning, specifically convolutional neural networks, in identifying Android malware.

Main Methods:

• Implementation of AMDDLmodel, a deep learning model utilizing a convolutional neural network (CNN).
• The model's performance is tuned using various parameters including filter sizes, epochs, learning rates, and network layers.
• Evaluation of the model using the Drebin dataset, which comprises 215 distinct features.

Main Results:

• AMDDLmodel achieved a high accuracy of 99.92% in detecting and classifying Android malware.
• The model demonstrated strong performance metrics, including precision, recall, and F1-score.
• Comparative analysis showed AMDDLmodel outperformed existing techniques in accuracy for Android malware detection.

Conclusions:

• AMDDLmodel represents an innovative deep learning solution for Android malware detection.
• The model significantly enhances detection accuracy and user security through advanced feature engineering.
• The findings highlight the potential of deep learning for robust mobile security and privacy protection.