AI RESEARCH
CoLA: A Choice Leakage Attack Framework to Expose Privacy Risks in Subset Training
arXiv CS.CV
•
ArXi:2604.12342v1 Announce Type: cross