Unveiling the Shadows: DarkNET Traffic Detection Using Deep Learning Techniques

In recent years, with the expansion of Internet usage and web browsing, a part of the public network has emerged as the “darknet”, which is an environment for illegal activities. To identify and classify this type of network traffic, machine learning and deep learning models are used in intrusion detection systems (IDS). This paper proposes a deep learning-Bayes estimator to build a DarkNET traffic classification model, with the aim of finding the architecture of the model in traffic discrimination. The method uses the Bayes estimator algorithm to search for the optimal hyperparameter configuration of the deep learning model, thereby improving the discrimination accuracy. The deep learning Bayes estimator achieves an accuracy of 98.52% when the model is applied to the CIC-Darknet2020 dataset. Although the proposed method effectively improves the accuracy of the model, it requires longer training time during the hyperparameter finding and initial model creation stage.