Improvement of Network Intrusion Detection Accuracy by using Restricted Boltzmann Machine

Abstract

In various data of network intrusion detection used for classification algorithm's learning, a great deal of noise and outlier data are mixed. In case of a learning performed by using data of high impurities, no matter how the performance of classification algorithm is outstanding, any network intrusion detection model of high performance becomes hard to anticipate. To increase the accuracy of network intrusion detection, not only the performance of classification algorithm should be increased but also the management on noises and outliers in the data used for the classification algorithm's learning. Restricted Boltzmann Machine (RBM) is a type of unsupervised learning that doesn't use class labels. RBM is a probabilistic generative model that composes new data on input data based on the trained probability. The new data composed through RBM show that the noises and outliers are removed from the input data. When the newly composed data are applied to the network intrusion detection model, negative effects from the noise and outlier data to the learning are eliminated. In this study, noises and outliers in KDD Cup 1999 Data are removed by applying the data to RBM and composing a new data. Then, use results between the existing data and the data from which noises and outliers are removed are compared. In conclusion, this study demonstrates the performance improvement of network intrusion detection resulted by removing noises and outliers included in the data through RBM.