College of Computer Science, University of Technology discussed a Ph.D. thesis on mechanisms for predicting cyberattacks based on an intelligent framework by the postgraduate, Ms. Azhar Falih Hassan.

The thesis reviewed developing an integrated and effective framework for detecting cyberattacks, with a focus on overcoming the major challenges facing current detection systems.

The thesis included presenting an advanced model for detecting multi-class cyberattacks using the Variational Autoencoder algorithm to solve the problem of detecting the most dangerous types of attacks and a cyberattack detection system based on semi-supervised technology and the Sparse Autoencoder algorithm, as well as creating a hybrid and integrated model that combines three important techniques (CNN, LSTM, and XGBoost) to predict cyberattacks.

The thesis reviewed a significant improvement in overall detection performance metrics, making it a promising solution for real-world applications of detecting and predicting various cyberattacks, while providing effective models for predicting the most complex and dangerous types of attacks with very high accuracy.