Improved traffic detection with support vector machine based on restricted Boltzmann machine

We can obtain a great deal of information from networks, but at the same time, we also face increasingly more problems, including those related to network security. Detecting network anomalies by their generation applications plays an important role in network security, and the quality of these systems is strongly dependent on the employed detection algorithms. Therefore, improving the performance of these algorithms is an important issue. In this paper, we design a new algorithm that we called the suppor vector machine based on the restricted Boltzmann machine (SVM-RBM) to detect network anomalies. The challenges for this algorithm are feature pre-processing and the speed for training the model. We use unsupervised algorithms such as the restricted Boltzmann machine (RBM) to extract useful features from the data sets and choose the gradient descent algorithm with Spark to train the support vector machine (SVM) classifier for short running time. Moreover, we explore the number of hidden units to improve the performance of SVM-RBM. We also discover that the learning rate has an effect on the SVM and we should choose the appropriate value.