Machine learning models and dimensionality reduction for improving the Android malware detection.

Area of Science:

• Computer Science
• Cybersecurity
• Machine Learning

Background:

• Android's widespread use presents numerous cybersecurity vulnerabilities.
• Existing Android malware analysis often relies on limited machine learning approaches.
• The DREBIN project provides a foundational dataset for Android malware research.

Purpose of the Study:

• To develop an efficient dimensionality reduction technique for Android malware features.
• To evaluate various supervised machine learning algorithms for malware prediction.
• To improve the accuracy and efficiency of Android malware detection systems.

Main Methods:

• Utilized the DREBIN dataset for Android malware analysis.
• Implemented an efficient feature dimensionality reduction method.
• Applied and compared multiple supervised machine learning algorithms, including Random Forest.

Main Results:

• Random Forest models demonstrated superior performance in malware detection.
• Achieved an average malware detection rate of 91.72% with a 0.13% false positive rate.
• Reduced feature set to 5,000 (9% of DREBIN features) while maintaining high accuracy (99.52%) and F1-score (96.99%).

Conclusions:

• Efficient dimensionality reduction combined with Random Forest offers a highly effective approach to Android malware detection.
• The proposed method significantly improves detection rates and reduces false positives compared to previous methods.
• This research contributes to more robust and efficient mobile security solutions.