An Effective Cognitive Cybersecurity System of Cognitive Sciences Using Adaptive Transformer-Based Deep Learning Model

Abstract

Based on the cybersecurity perspectives, cognitive sciences are useful to improve human factor capabilities, especially in complex systems developed by technologies like social networks, Internet of Things (IoT), mobile, and cloud that generate vast amounts of data. In recent days, it is more important for scientists to analyze how cognitive sciences are helpful in improving human capabilities for cybersecurity tasks. The current cognitive security models aim to improve cybersecurity tasks by integrating some solutions. Nevertheless, they still have problems related to the intrinsic shortcomings of each component. Hence, by utilizing cognitive science, a novel cognitive cybersecurity system is developed in this work. The required data are garnered initially from diverse online sources. Then, the gathered data is fed as input to the Adaptive Transformer-based Variational Autoencoder with Long Short-Term Memory (AT-VAELSTM) for decision-making purposes. Here, the decision-making process is improved by optimizing the AT-VAELSTM model parameters with the assistance of the Renovated Position-based Pine Cone Optimization Algorithm (RP-PCOA). Finally, an efficient outcome is achieved from this model, and the necessary experimental analysis is performed for the designed model by comparing it with the baseline approaches to guarantee the supremacy of the model.