Quantum transfer learning for cross-domain cybersecurity threat detection and categorization

Cybersecurity challenges have become increasingly complex and widespread, and the risks associated with these problems are substantial, affecting thousands of individuals and organisations and being crucial to national security. As cybercriminals have become increasingly adept at utilizing advanced methods to exploit system vulnerabilities, there has never been a more pressing need for reliable threat detection and response systems. This study proposes a framework that uses quantum transfer learning to enhance cybersecurity threat detection by leveraging multiple datasets, including UNSW-NB15, CICIDS2017, CSE-CIC-IDS2018, and TONIoT. The framework focuses on improving the accuracy and efficiency of existing machine learning methods for cyber threat detection by employing quantum computing techniques for feature extraction and analysis. The pre-processing of the UNSW-NB15 dataset, the extraction of quantum features using PennyLane, and the training of the deep learning model with TensorFlow are the steps in the workflow of this study. Finally, the model is fine-tuned through transfer learning on other datasets, resulting in improvements in detection accuracy. This study shows that our quantum-enhanced model attains an accuracy of 83% on UNSW-NB15, 91% on the combined CICIDS2017 and CSE-CIC-IDS2018 datasets, and 86% on the TONIoT dataset, demonstrating the potential of quantum computing and its use in the field of cybersecurity. Unlike fully quantum classifiers, our approach applies quantum transformations only at the feature-extraction stage, thereby creating a hybrid classical-quantum workflow that enhances transfer-learning performance across multiple cybersecurity datasets.