1 paper · 1 filter
Christina Runkel, Kanchana Vaishnavi Gandikota, Jonas Geiping +2
Being able to reconstruct training data from the parameters of a neural network is a major privacy concern. Previous works have shown that reconstructing training data, under certa…