Tag: machine learning security
Browse our exclusive articles!
Semantics-Aware Backdoor Attacks in Federated Learning
Explore SABLE, a novel semantics-aware backdoor attack in federated learning that uses natural triggers to evade detection and compromise model integrity.
SNEAKDOOR: Stealthy Backdoor Attacks on Dataset Condensation
Discover SNEAKDOOR, a stealthy backdoor attack method targeting distribution matching-based dataset condensation with high effectiveness and invisibility.
Transfer of Adversarial Robustness Across Perturbation Types
Explore how adversarial robustness transfers between perturbation types to enhance AI model resilience and improve training strategies.
Enhancing AI Robustness Against Unforeseen Adversaries
Discover how Unforeseen Attack Robustness (UAR) improves AI model resilience against unexpected adversarial attacks for secure real-world applications.