An Android Malware Detection System Based on a Hybrid Artificial Neural Network and Decision Tree
Oforjetu Chukwudi Peter *
Department of Computer Science, Federal University Wukari, Nigeria.
Andrew Ishaku Wreford
Department of Computer Science, Federal University Wukari, Nigeria.
*Author to whom correspondence should be addressed.
Abstract
When it comes to sharing data, the broad use of cloud computing has been a huge boon. The development of cloud computing has brought many benefits, but it has also opened the door for criminals to steal sensitive information by taking advantage of its accessibility. This research looks at how deep learning and machine learning techniques, particularly ANN and DT algorithms, can be used to counteract malware. To construct a hybrid model, these methods are combined using a stacking methodology. The investigation was conducted using the Android Network Traffic dataset from Kaggle. We use the Information Gain algorithm for feature selection and a variety of metrics to measure the models' performance, with accuracy serving as the main indicator. On the Android network traffic dataset, the hybrid model attained a 99% accuracy score, according to the results. To improve data security in cloud-based systems and deal with malware, this study shows that integrating deep learning and machine learning approaches works
Keywords: Anomaly-based, malicious software, signature-based, android malware, heuristic-based detection