Enhancing Artificial Neural Networks for Network Based IDS

摘要

Network-based intrusion detection model using Artificial Neural Networks (ANN) for detection is presented in this paper. The key idea is to take advantage of classification abilities of neural network in the identification of external attacks against a network. By a careful analysis of network data the new model has ability to recognize attacks, differentiate one attack from another i.e. classify and identify the attack, and detects novel attacks with high detection rate and low false alarms, based on limited, incomplete, and nonlinear data sources. New Neural Network training approach is implemented in this IDS proposal, which minimize training time and error rate for new presented data.