The project, titled "An Effective Temporal Convolutional Networks Based Method for Detecting Android Malware Using Dynamic Extracted Features," aims to build an intelligent and robust malware detection system using dynamically generated behavioral features from Android applications. The system targets classification of apps into five categories: Adware, Banking Malware, SMS Malware, Riskware, and Benign. To achieve high detection accuracy, the system leverages a diverse set of machine learning and deep learning models including existing algorithms such as Deep Neural Network (DNN), LightGBM, Convolutional Neural Network (CNN), Multi-Layer Perceptron (MLP), and particularly the Temporal Convolutional Network (TCN), which captures sequential dependencies in app behavior.To enhance detection performance and generalization, the project introduces advanced ensemble techniques including a Stacking Classifier, Voting Classifier, and a hybrid Autoencoder-based Deep Neural Network, where an autoencoder is used for effective feature compression followed by an MLP classifier for final prediction. This layered architecture improves malware discrimination by learning rich latent patterns in dynamic behavior data. The system is trained on a labeled dataset with feature-rich dynamic traces and evaluated across all five malware types to identify the best-performing model.By combining time-series modeling, deep learning, and ensemble learning, this system provides a comprehensive and scalable solution for Android malware classification. It supports proactive threat detection, thereby enhancing mobile security ecosystems.

Keywords: Android Malware Detection, Temporal Convolutional Network (TCN), Autoencoder, Dynamic Analysis, Ensemble Learning, Deep Neural Network, Stacking Classifier, Voting Classifier, Behavioral Features, Mobile Security.