Cybersecurity #cybersecurity#cyberattack
GDGU Method Speeds Cyberattack Localization in EV Charging Networks by 10x While Preserving Data Privacy
Researchers propose GDGU, a gradient difference-based graph unlearning method for cyberattack localization in electric vehicle charging networks. GDGU removes the influence of deleted training data without full retraining, achieving 10-12x speed improvement while maintaining localization accuracy close to retraining. The method addresses privacy regulations that allow EVCS owners to delete data from deployed models.
Jun 20, 2026 1 source