SCADA intrusion detection system based on self-learning Semi-Supervised One-Class Support Vector Machine
Xiaojun Yang · Metallurgical Industry Automation · 2013
In order to study the security issues in industrial control systems from the perspective of manufacturing process,a SCADA intrusion detection system based on self-learning Semi-Supervised One-Class Support Vector Machine(S2 OCSVM) is designed.Since the SCADA system data are of smaller sample size and higher dimension,S2 OCSVM algorithm is adopted to construct the classifier.By designing the active learner,the system can be able to add representative samples capable of improving the performance of the classifier into the training set,so as to increase the classifying accuracy,in other words,reduce the rate of false-alarm and miss-alarm.The experimental results show that the present method can effectively improve the detection accuracy,but the real-time of the classifier training needs to be further enhanced.