ZAITANG LI

certified robustness (1) adversarial robustness (1) model evaluation (1) adversarial attack (1) generative model (1) adversarial perturbation (1) generative adversarial network (1) vision-language model (1) global robustness (1) model ranking (1) jailbreak attack (1) robustness evaluation (1) multi-modal evaluation (1) retention score (1) toxicity score (1) certified perturbation (1) perturbation evaluation (1) certified attack (1)