Intrusion Detection Basing on Incremental Support Vector Machine
Yu Gu · Journal of Yunnan University for Nationalites · 2005
New intrusion behaviors emerge endlessly in computer network. The intrusion detection systems are required to possess the ability to learn new types of intrusion. This paper designs an incremental SVM (support vector machine) training algorithm based on survival factor. This algorithm represents the known intrusion information not only by the marginal samples, but also by the quasi-marginal samples. So it can learn new intrusion information in an incremental form. Compared with the pure SVM algorithm, the incremental training SVM algorithm based on survival factor can suppress the vibration phenomenon occurred in the training process, and improve the adaptiveness and robustness of the algorithm effectively.