1 paper · 1 filter
Viktor Valadi, Mattias à kesson, Johan Ãstman +3
Gradient inversion attacks are often presented as a serious privacy threat in federated learning, with recent work reporting increasingly strong reconstructions under favorable exp…