Anomaly Detection Using XGBoost Ensemble of Deep Neural Network Models

Abstract Intrusion Detection Systems (IDSs) utilise deep learning techniques to identify intrusions with maximum accuracy and reduce false alarm rates. The feature extraction is also automated in these techniques. In this paper, an ensemble of different Deep Neural Network (DNN) models like MultiLayer Perceptron (MLP), BackPropagation (BPN) Long Short Term Memory (LSTM) are stacked build a robust anomaly detection model. performance the model analysed on datasets, namely UNSW-NB15 campus generated dataset named VIT_SPARC20. Other types traffic, unencrypted normal encrypted malicious captured VIT_SPARC20 dataset. Encrypted traffic categorised by without decrypting its contents, thus preserving confidentiality integrity data transmitted. XGBoost integrates results each achieve higher accuracy. From experimental analysis, it inferred that UNSW_ NB maximal 99.5%. terms accuracy, precision recall 99.4%. 98% 97%, respectively.