AI RESEARCH

CoLA: A Choice Leakage Attack Framework to Expose Privacy Risks in Subset Training

arXiv CS.CV

ArXi:2604.12342v1 Announce Type: cross