Android Malware Detection Using Machine Learning and Neural Network: A Hybrid Approach with Federated Learning

Android is the most popular mobile operating system, which makes it a major target for malware attacks. This paper focuses on the growing problem of malware targeting Android devices and presents a new approach for detecting such threats. The proposed method combines machine learning, neural network and Federated Learning to improve malware detection. A majority voting technique is used to select the most important features from the dataset, which helps identify harmful behaviors more effectively. Various machine learning algorithms, including Random Forest and XGBoost, and neural network architectures, such as CNN-LSTM, are evaluated on an Android malware dataset. Federated Learning ensures data privacy during training, addressing key cybersecurity concerns. The study emphasizes the importance of advanced techniques for adaptive malware detection and provides a detailed analysis of performance across multiple models. Results indicate high accuracy, underscoring the efficacy of the proposed methods in combating Android-specific malware.