Comparative Study of Machine Learning Algorithms for Network Traffic Classification
Traffic Recognition is the most arduous task. With the expansion of cyberspace, substantial number of people are using internet every day. It has become the important part of our life. People exchange their personal or professional information on internet, so it is very important to safeguard their information from an unauthorized person. Although there are many traffic recognition techniques that were developed in the past which were effective back then but with the time, we need some more effective technique to differentiate between malicious or non-malicious data packets. Machine learning provide enormous number of data classification technique. In this paper, we analysed and compared some of the machine learning algorithmson NSL-KDD dataset using Weka toolfor traffic recognition. Our experimental process consists of three steps. In the first step, we compare existingMachine Learningalgorithmswithout any feature selection algorithm. Then in the second step,some of the feature selection algorithms are applied and finally over-sampling and under-sampling algorithms are used for comparative study.