An Intrusion Detection Framework with Optimized Feature Selection and Classification Combination Using Support Vector Machine

As intelligent devices and related technologies continue to develop, abnormal traffic detection has become a major issue in the field of internet security. Malicious attacks can have a negative impact on systems and cause a decline in computational performance. One of the technologies used to identify intruder activity and assess system security by sending out notifications is intrusion detection systems. In this endeavor, we employ the diversified NSL-KDD dataset for intrusion detection to introduce a novel feature selection and classification merging technique using support vector machines (SVM). The goal of this technique is to increase the capability of intrusion classification by considerably reducing the input feature set from the training data. The process of choosing significant input training features and discarding unimportant ones in supervised learning results in a feature subset that can produce greater classification accuracy. In our tests, we used the KDDTest+ and KDDTest-21 data sets’ various input feature subsets as input features for the SVM classifier. According to the experimental findings, the suggested technique obtains classification accuracy of 88.25% when using only KDDTest+ and 72.42% when using KDDTest-21.