IDS FEATURE REDUCTION USING TWO ALGORITHMS

In recent years, intrusion is defined as detection of any security threats .The security of the information has become a very dangerous problem in the security of the data and network. Highly secret data of different arrangements are present via the network so in order to protect that data from unauthorized users, it is required a very strong security structure. An IDS (Intrusion detection system) gathers and tests information from various areas within a network to determine the most likely security threats that from both outside and inside the system. IDS deals with huge data which include different redundant and irrelevant features that results in increasing time processing and decreasing detection rate. Therefore reduction of features plays an important role in IDS. In this paper two dimensionality reduction algorithms PCA and SVD were implemented on KDDCUP'99 dataset. Experimental results were obtained to get the best reduced feature set that recognized using SVM algorithm. Detection rate, error and accuracy are used to evaluate IDS Performance.