An Efficient Network Intrusion Detection System using Multilayer Perceptron

Presently, the growth of internet has given rise to increasingly utilize internet for public and commercial services. As a result malicious activities are on the rise to exploit users to steal information related to credit card details, passwords, sensitive information etc. It has become a major concern in the society to protect data shared across networks and stored in computer. At the same time attackers use many techniques to gain advantage of breaking firewalls and security measures. The attacker use techniques to steal the data such as malware, viruses, Trojans etc. An intrusion detection system is used to detect and block malicious traffic originating from attackers. Traditional Intrusion detection systems are becoming less effective to capture the malicious traffic. Combining data mining techniques into intrusion detection system offer more security than traditional methods. Since data mining can use vast number of data, detecting a tiny variation in the network traffic is possible leading to detect and block unwanted traffic in the network. In the present study multi layer perceptron classifier is used to classify the network attack types. MLP based intrusion detection system is designed to support network administrators to find signature based attacks and anomaly of the network traffic. The IDS is trained on the traffic data using MLP to capture the known attack types and differentiate traffic as intrusion and normal traffic.