Network Intrusion Detection System Using KNN and Naive Bayes Classifiers

Abstract

For every second after digitization, enormous quantities of data are generated from different networks. The value of providing this data with protection has therefore increased. The need to automate this protection framework has become necessary as the data is really massive. Intrusion detection systems are known as a great approach for intrusion detection. The system for detecting intrusions serves as an important mechanism to detect web security attacks. A proven technique for detecting network-based attacks, the Intrusion Detection System is still inexperienced in monitoring and recognising attacks, but efficiency remains unchanged. A large number of techniques that are based on machine learning approaches have been developed.

In this paper, the CIDDS-001 dataset [12] is analyzed and the observations have been marked. Two supervised machine learning techniques such as K-Nearest Neighbour and Naive Bayes are implemented on this dataset. KNN is implemented with different K – values. KNN is executed by changing the number of testing records. K – Value is also decided by making keen observations. KNN algorithm gives on an average 92.3% accuracy. Naive Bayes algorithm is also executed by changing the number of testing records. Naive Bayes algorithm gives on average 70.66% accuracy. Time complexity of NB algorithm is less than KNN. Comparison of both the methods is presented by comparing their evaluation metrics like Accuracy, Precision, Recall, Specificity and F- Measure. This paper is concluded by identifying the pros and cons of both the algorithms and by providing the future scope of this paper.