AI RESEARCH
Revisiting the LiRA Membership Inference Attack Under Realistic Assumptions
arXiv CS.LG
•
ArXi:2603.07567v1 Announce Type: cross Membership inference attacks (MIAs) have become the standard tool for evaluating privacy leakage in machine learning (ML). Among them, the Likelihood-Ratio Attack (LiRA) is widely regarded as the state of the art when sufficient shadow models are available. However, prior evaluations have often overstated the effectiveness of LiRA by attacking models overconfident on their