ABSTRACT:

Intrusion detection faces a variety of challenges; an intrusion detection system must reliably detect malicious activities in a very network and should perform efficiently to cope with the big amount of network traffic. In this paper, we address these 2 issues of Accuracy and Potency using Conditional Random Fields and Layered Approach. We have a tendency to demonstrate that high attack detection accuracy will be achieved by using Conditional Random Fields and high efficiency by implementing the Layered Approach. Experimental results on the benchmark KDD ’99 intrusion information set show that our proposed system based mostly on Layered Conditional Random Fields outperforms other well-known strategies like the decision trees and also the naive Bayes. The improvement in attack detection accuracy is terribly high, particularly, for the U2R attacks (thirty four.8 percent improvement) and also the R2L attacks (thirty four.five percent improvement). Statistical Tests also demonstrate higher confidence in detection accuracy for our methodology. Finally, we tend to show that our system is strong and is able to handle noisy data while not compromising performance.