A self-evolving AI-based Intrusion Detection System (AI-IDS) using Random Forest and Variational Autoencoder (VAE) for hybrid detection of known attacks and zero-day malware, with a real-time Human-in-the-Loop feedback mechanism for continuous model retraining.

Official PyTorch implementation of ICS-PhysGAN: Generating High-Fidelity Zero-Day Industrial Attacks using Physics-Constrained Adversarial Learning (WGAN-GP) on the SWaT dataset.