MixNN: A design for protecting deep learning models

Chao Liu, Hao Chen, Yusen Wu +1

In this paper, we propose a novel design, called MixNN, for protecting deep learning model structure and parameters. The layers in a deep learning model of MixNN are fully decentra…