Research of Intrusion Detection Algorithm Based on Parallel SVM on Spark

As network intrusion data's scale gets larger and larger, designing parallel schemes for intrusion detection have been becoming research focus in the field of information security. In order to solve the problem that the intrusion detection algorithm is high time-consuming, the classification of large amounts of data occupies lots of memory and the efficiency of single detection is low, a parallel principal component analysis (PCA) combined support vector machine (SVM) algorithm based on spark platform is proposed (SP-PCA-SVM) in this paper. This method adopts the way of principal component analysis (PCA) for training and predicting data, and then introduces a fusion of Bagging integration strategy and SVM algorithm, and finally uses spark distributed framework to achieve. The results show that, for a large number of intrusion data, the new parallel training method, to a certain extent, reduces the training time and improves model learning efficiency.