An Effective Intrusion Detection System using Supervised Machine Learning Techniques

With the increased use of Internet resources, cyber attackers are using novel ways to attack the services of network. Thus network security is becoming inevitable part of the network system. In order to detect such attacks efficiently and effectively, robust IDS (Intrusion Detection System) is needed. An IDS is a tool that analyzes each and every packet deeply to detects malicious activity by monitoring a network or a system. The main purpose of IDS is to identify unwanted or abnormal action and to inform the network administrator about such actions. Thus, IDS is important tool for the network administrator to prevent the network from both known and unknown attacks that make the network resources more vulnerable. Machine learning methods can be used to employ efficient intrusion detection system (IDS). In this research work four machine learning methods were used namely RF (Random Forest), DT (Decision Tree), MLP (Multilayer perceptron) and SVM (Support Vector Machine) for classification of the data. NSL-KDD dataset was used for training and testing these various machine learning models. Feature selections were used to eliminate the irrelevant and unwanted features from the dataset. Therefore feature selection reduces the dimensionality of the dataset which in turn reduces the computational complexity. The proposed model's output was evaluated using three feature subsets, randomly selected from the NSL-KDD dataset. The proposed method has a classification accuracy of more than 99 percent.