Intrusion detection system based on growing grid neural network

The use of neural networks in the area of intrusion detection systems has significantly increased over the last few years. In this paper, we present the results obtained by comparing the Growing Grid neural network and the Self-Organizing Maps applied to the intrusion detection systems. We compare two important aspects, the performance and the training time. The results show that the increasing network improves the performance of the system in detection of anomalies obtaining better relation between the detection rate and the number of false positives. On the other hand, a very significant reduction of the training time in real environments is obtained. The networks have been trained and tested with data provided by the DARPA Intrusion Detection Evaluation program.