Network Traffic Threat Feature Recognition Based on a Convolutional Neural Network
Yang Gao, Anilkumar Kothalil Gopalakrishnan · 2019
This paper presents a novel algorithm for recognizing threats from communication network traffic features based on a Convolution Neural Network (CNN). The CNN extracts higher-dimensional network features from the network traffic dataset by deepening the number of its convolution layers. The transfer learning method with Principal Component Analysis (PCA) is applied for the CNN training. The transfer learning method trains the CNN initially with a small amount of dataset, and then it trains the network with the complete dataset. The PCA is a technique for analyzing, and simplifying dataset, and it often used to reduce the dimensionality of the dataset. The experimental results showed that the presented system could be an effective way for recognizing threats from a communication network.