ABSTRACT
The rapid evolvement of advanced technology is driving the world’s economic landscape. Every single sector is directly or indirectly attached with this fourth industrial revolution. Thereby, the scope for fraudulent entities is also continually enlarging while cyber attack is the main threat. Detecting traffic and accessing devices are the main focus of this study. Therefore, this chapter is conducted to determine the best classifier for detecting the access of the DarkNet using tree-based, distance-based, probability-based, and basic deep learning algorithms. Furthermore, this study is intended to identify the precise dimensionality reduction and data balancing approaches for cyber security. The result reveals that undersampling of the data may result in overfitting, which is suggested to avoid in this domain. This study suggests the ICA and SMOTETomek pipelines as the pre-processing techniques to classify the DarkNet from CICDarknet2020. This chapter opens up a new debate for cybersecurity scholars to determine the best way of identifying fraudulent individuals.
