A Novel User Membership Leakage Attack in Collaborative Deep Learning
WCSP, pp. 1-6, 2019.
Collaborative deep learning can provide high learning accuracy even participanted usersu0027 datasets are small. In the training process, users only share their locally obtained parameters, therefore it is believed that the privacy of usersu0027 original datasets can be protected. However, we present an attack approach against usersu0027 ...More
Full Text (Upload PDF)
PPT (Upload PPT)