Network Traffic Image Dataset Generation from PCAP files for Evaluating Performance of Machine Learning Models

Detection of network attack traffic in network environments is majorly studied in the literature by applying various data mining and machine learning techniques. The existing studies which applied data and machine learning techniques consider network traffic instances in either pcap or csv representations. In the current contribution, the basic idea is to use network traffic images which are obtained from pcap representations. These generated network traffic images are then used to design and build efficient machine learning models. Another advantage of representing the network traffic in the form of images is that these images can be used to evaluate the computational performance of deep learning models also. Till date, the existing machine learning studies on cloud network traffic attack detection and network environments did not apply network traffic images to build machine learning models but only applied to build deep learning models which form the motivation for the present research. We propose to use the network traffic images for designing new machine learning models.