Akademska digitalna zbirka SLovenije - logo
E-viri
Celotno besedilo
  • Liao, Zhiyong; Luo, Junren; Gao, Wei; Zhang, Yu; Zhang, Wanpeng

    2019 Chinese Automation Congress (CAC), 2019-Nov.
    Conference Proceeding

    Homomorphic encryption assisted privacy preserving learning has become one powerful technology in actual scenarios and applications. In this paper, we design a novel homomorphic CNN in conjunction with homomorphic encryption to make prediction on encrypted sensor data. We carry out privacy preserving learning experiments with different activation functions and evaluate the accuracy on the MSTAR dataset. Experimental results show the models' effectiveness on encrypted sensor data with privacy preserving.