An Improvement of Neural Network Algorithm for Anomaly Intrusion Detection System
Mohammed Qassim Shatnawi, Muneer Bani Yassein, Shadi A Aljawarneh, Sawsan Alodibat, Omar Meqdadi, Ismail Hmeidi, Omar Al Zoubi · International Journal on Communications Antenna and Propagation (IRECAP) · 2020
In this study, the focus has been on analyzing various types of network attacks using machine learning algorithms. A model based on Neural Networks and Deep Learning for detecting network attacks has been evaluated. The proposed model has been tested on CICIDS2017 dataset. The performance of the presented algorithm is compared against a set of other machine learning algorithms; k-Nearest Neighbors, Interactive Dichotomizer 3, Random Forest, Naïve Bayes, Adaboost, Quadratic Discriminant Analysis, and Multiplayer Perceptron. Results have showed that the adopted neural network algorithm presents a better accuracy of detecting different types of network attacks. An accuracy of up to 99% has been attainted, and false negative alarms have been reduced. Finally, it has been found out that by using a deep learning approach time of detection is reduced to half of time.