Study of different malicious attacks with data mining techniques and its extraction of knowledge using various machine learning models for intrusion detection

Abstract

The rise of online applications, such as online shopping, hospital record newlinestorage, educational course, money transfer, infotainments, etc., has increased the use newlineof interest, which has simplified the work. The rise in internet usage also tends to newlinecause more malicious attacks Protection mechanisms such as firewalls and antivirus newlinedetect most of the malicious attacks. There are few unidentified malicious attacks newlinewhich can cause severe loss of valuable or personal information. The fast-growing newlinecomputer and wireless networks are vulnerable to malicious violence. This problem is newlinesolved by intrusion detection system. Intrusion Detection System (IDS) plays a vital newlinerole in shielding data from malicious attacks on computer networks. T he main newlinechallenge for the development of an IDS model is to optimize accuracy, predict time newlineand increase the false negative rate. Existing IDS model methods have disadvantages newlinesuch as higher prediction time, higher false negative rate, lower recall and less newlineaccuracy. The IDS approach is best suited for the development of the IDS platform, newlinebased on data mining and machine learning methods. newline10% Kddcup 99 dataset is different percentages considering all possible attacks was newlineutilized, preprocessing was performed using one hot encoding and we proposed newlinestatistical methods are used obtain Z-scores, mean, median and mode for determining newlinethe significant features, training was performed for 80% of the dataset. The remaining newline20% dataset was tested and validation using decision tree, K-NN and Gradient newlineboosting algorithm. The efficiency of these algorithms was analyzed and discussed it newlinealso found that Gradient boosting algorithm is best suitable algorithm to be utilized newlinefor development of IDS. Different parameters were interacted based on the testing and newlineare false negative rate, accuracy, f-score, precision time. These metrics are vulnerable newlineto the efficiency of the IDS model that was developed. newlineThe proposed Gradient Boosting IDS model achieving prediction of accuracy newlinefor all attacks is 99.58 percent, fal