To search, Click below search items.


All Published Papers Search Service


Network Traffic Vulnerability Analysis using Machine Learning- A comparative approach


Shrabani Mallick Dharmender Singh Kushwaha


Vol. 20  No. 6  pp. 28-35


With the increase in use of web application to enable business and social networking rapid application development and deployment has been commonplace, this has increased the risk of potential network threats vulnerabilities. Thus, one the biggest challenge of the time is insecure codes running on various servers in the network making the network vulnerable and susceptible to network security breaches. Various Machine Learning using supervised and unsupervised models have been widely used to delve deep into network access log data to discover network vulnerabilities. This paper presents a comparative study of 3 machine learning approaches Na?ve Bayes, k-Nearest Neighbour (kNN) and an Artificial Neural Network (ANN) to analyze the network access logs for vulnerability, particularly involving application access over network. The results are quite convincing with Na?ve Bayes model with an accuracy score of 94% as compared to K- Nearest Neighbour with an accuracy of 85.2%. MLP is also reckons an accuracy score of 90.37% with very high prediction rates. The training times of MLP is of course high due to the number of epochs.


Na?ve Bayes, kNN, ANN, Supervised, Unsupervised learning