Enhanced Big Data in Intrusion Detection System by Machine Learning

Abstract

In the last years, a number of peoples around the world are increasing day by day to use the internet and social media. For this reason, a huge volume of information is generated over the internet and social media from gigabytes (GB) to peta-bytes (PB) with high speed. The 3vs model of Big Data which runs into trillions and millions of items and bits of information. Basically, the 3Vs (velocity, volume and variety) are the three characteristics and dimension of Big Data. Volume defines the amount of information, variety defines various types of data in numbers and velocity defines data processing speed.  The Principal Component Analysis (PCA) is applied to reduce data dimension while preserving information of original data and building intrusion detection models by using K-nearest neighbors (K-NN) and Support Vector Machine (SVM) classifier to evaluate the accuracy of the system. The researcher has applied KDD99 datasets to train and test the model. Thus, the researcher has introduced a comparison between the K-NN classifier and SVM classifier. The results of the experiment showed that the SVM classifier high performance, which reduces the false positive rate as well as giving high accuracy and is efficient for Big Data.