AI and Machine Learning Models for Real-Time Intrusion Detection Systems

Real-time intrusion detection systems (IDS) are critical for safeguarding modern networks against evolving cyber threats. Artificial intelligence (AI) and machine learning (ML) models significantly enhance IDS capabilities through anomaly detection, predictive analytics, and automated response mechanisms. This paper reviews state-of-the-art AI/ML models for real-time IDS, including supervised, unsupervised, and deep learning techniques such as XGBoost, convolutional neural networks (CNNs), long short-term memory networks (LSTMs), and hybrid architectures. Drawing from 2025 literature, it examines applications in Internet of Things (IoT) and enterprise network environments, evaluates performance on benchmark datasets including NSL-KDD, UNSW-NB15, and CICIDS2017, and explores integration with explainable AI (XAI) for model transparency and trust. Key findings highlight optimized models achieving detection accuracies exceeding 99%, while addressing challenges such as computational overhead, scalability, and adversarial robustness. A graphical comparison of model performances is presented. Future research trends emphasize edge AI deployment and quantum-resistant security frameworks for resilient, scalable, and real-time intrusion detection.