A Robust and Efficient Machine Learning Framework for Enhancing Early Detection of Android Malware

The advancement of information technology has introduced new challenges in cybersecurity, especially related to the Android platform which is the main target of malicious software (malware) attacks. The National Cyber and Crypto Agency (BSSN) of Indonesia reported millions of incidents involving Android Package Kit (.apk) files related to electronic wedding invitations. This study aims to develop a robust and efficient static analysis-based machine learning framework for early detection of Android malware. Six machine learning algorithms Support Vector Machine (SVM), Random Forest (RF), K-Nearest Neighbors (KNN), Naive Bayes, AdaBoost, and Gradient Boosting are used to identify malicious behavior in APK files. The dataset used consists of 2,084 Android applications, including 1,314 malware samples and 770 benign applications, obtained through a reverse engineering process. Data pre-processing, feature extraction, and training using supervised learning are carried out to optimize detection accuracy. The experimental results show that the Random Forest algorithm achieves the best performance with 97% accuracy and 95% precision, surpassing the performance of other algorithms.