Building an Intrusion Detection System to Detect Atypical Cyberattack Flows

Artificial Intelligence (AI) techniques provide effective solutions for the detection of many aberrant network traffic patterns and attack flows. However, validation these often relies on one training dataset. Recent results show that such may fail in face dynamically-changing cyberattacks. Given increased sophistication cyberattacks nowadays, it is imperative to examine improve performance AI models. This paper proposes a defensive engine combined with twofold feature selection technique hyperparameter optimization model. In this work, we utilize proposed system binary flow identification models are trained validated CICIDS2017 The then evaluated using synthesized atypical flows mimic real-world scenarios. We demonstrate effectiveness approach several Deep Learning Machine including DNN, Linear-SVC, Stacked Decision Tree Classifier (S-DTC). Simulation significantly improves True Positive Rate (TPR) multiple attacks.