A Neural Network Model for Detection Systems Based on Data Mining and False Errors

Nowadays, computer network systems play an increasingly important role in our society. They have become the target of a wide array of malicious attacks that can turn into actual intrusions. This is the reason why computer security has become an essential concern for network administrators. Intrusions can wreak havoc on LANs. And the time and cost to repair the damage can grow to extreme proportions. Instead of using passive measures to fix and patch security holes, it is more effective to adopt proactive measures against intrusions. Recently, several IDS have been proposed and they are based on various technologies. However, these techniques, which have been used in many systems, are useful only for detecting the existing patterns of intrusion. It can not detect new patterns of intrusion. Therefore, it is necessary to develop a new technology of IDS that can find new patterns of intrusion. This paper investigates the asymmetric costs of false errors to enhance the detection systems performance. The proposed method utilizes a network model considering the cost ratio of false errors. Compared with false positive, this scheme accomplishes both security and system performance objectives. The results of our empirical experiment show that the network model provides high accuracy in detection. In addition, the simulation results show that effectiveness of probe detection can be enhanced by considering the costs of false errors.