Set-level Guidance Attack: Boosting Adversarial Transferability of Vision-Language Pre-training Models | IEEE Conference Publication | IEEE Xplore