Hybrid Quantum-Classical Intrusion Detection Framework for Intelligent Cyber Threat Detection
The number of digital communications and cloud infrastructure is rising rapidly, cyber attacks have become much more advanced and frequent. Conventional approaches for intrusion detection cannot detect the evolving threats due to highly dimensional traffic patterns and complex attacker behavior. To tackle these issues, we introduce a hybrid quantum-classical intrusion detection approach by integrating classical and quantum machine learning algorithms to build an efficient intrusion detection system. Our model takes advantage of UNSW-NB15 Cybersecurity benchmark dataset that contains modern network attack vectors including exploit, denial of service attack, reconnaissance, shellcode, and backdoor attacks. For data preparation, we implement various data processing methods including categorical encoding, feature scaling, and dimensionality reduction. For comparison purposes, Random Forest algorithm is used as a classical detection method and variational quantum classifier based on Qiskit software library is utilized as our quantum learning model. Our experimental results show that our hybrid model is highly accurate in detecting the attacks. This work provides a comparative analysis between classical and quantum learning models demonstrating the promising future of quantum computers in cyber attacks detection tasks.