One-class intrusion detection technique based on Bayesian approach
Xiao Xianqia · Journal of Hebei University · 2014
To consider the imbalance of the sample data in intrusion detection problem,one-class models can be applied to intrusion detection.The one-class support vector machine(one-class SVM)was improved to be a probabilistic model by Bayesian approach,which makes it more conform to the random characteristics of intrusion process.The source data was preprocessed with the same variance in each orientation by applying principle component analysis(PCA)technique,which makes it to be more suitable to the data normality hypothesis of the model.The idea of partition was applied in solving the model to calculate each data packet,so as to realize the efficient solution of large data.Test the model with the standard intrusion detection dataset NSL-KDD,achieve the experiment result with 83.96% detection accuracy,which validate the effectiveness of the techniques in intrusion detection application.