Unknown Attack

Unknown attack detection focuses on developing robust systems capable of identifying and responding to previously unseen malicious activities, a critical challenge across diverse domains like cybersecurity and biometric authentication. Current research emphasizes one-class classification methods, often employing models like transformers and hyperbolic embeddings, to learn normal behavior and flag deviations as potential threats, supplementing these with techniques like synthetic data generation and clustering to improve generalization. This field is crucial for enhancing security in various applications, from protecting industrial control systems and IoT networks to securing face recognition systems against sophisticated spoofing attempts.