Black-Box Membership Inference Attack for LVLMs via Prior Knowledge-Calibrated Memory Probing

Re-assign community