Fusion of BVM and ELM for Anomaly Detection in Computer Networks
Changning Cai, Huaxian Pan, Guojian Cheng · 2012
This paper proposes a new network anomaly detection method in order to deal with the low detection rate and high false alarm rate problem. Ball vector machine (BVM) and extreme learning machine (ELM) is individually applied to learn three kinds of network features, then a BP neural network is utilized to simulate weights, which is used to fusion of the label. The experiments show that, the performance of this fusion method is better than single BVM or ELM classifier. Compared to the fusion method of SVM and BP neural network, the method proposed by this paper has a similar performance in detection rate and false alarm rate but with a significantly lower training time, and it is suitable for network anomaly detection with large scale dataset.