1 paper
Lukas Struppek, Dominik Hintersdorf, Antonio De Almeida Correia +2
Model inversion attacks (MIAs) aim to create synthetic images that reflect the class-wise characteristics from a target classifier's private training data by exploiting the model's…