TONIC: Towards oblivious neural inference compiler

Po Hsuan Huang, Chia Heng Tu, Shen Ming Chung

研究成果: Conference contribution


Privacy-preserving deep learning computing becomes popular these days as it helps protect, for example, both user data and deep neural network (DNN) model parameters at the same time with cryptographic techniques. In particular, significant efforts have been made to leverage secure two-party computation schemes for preventing user/model data from disclosure during DNN inference. Nevertheless, the existing works require manual intervention while converting trained models into secure computation programs, which is not scalable to modern deep networks efficiently. In this work, we propose a compiler framework, TONIC, to do the conversion automatically with scalability. Given a pre-trained DNN model, TONIC converts it into one of two secure two-party computation languages, i.e., ObliVM and ABY. Based on tailored backends built on top of a DNN compiler, TVM, our case studies show that TONIC is able to automatically convert popular DNN models, such as CryptoNets and MobileNetV2, into the corresponding programs for secure computations.

主出版物標題Proceedings of the 36th Annual ACM Symposium on Applied Computing, SAC 2021
發行者Association for Computing Machinery
出版狀態Published - 2021 三月 22
事件36th Annual ACM Symposium on Applied Computing, SAC 2021 - Virtual, Online, Korea, Republic of
持續時間: 2021 三月 222021 三月 26


名字Proceedings of the ACM Symposium on Applied Computing


Conference36th Annual ACM Symposium on Applied Computing, SAC 2021
國家/地區Korea, Republic of
城市Virtual, Online

All Science Journal Classification (ASJC) codes

  • 軟體


深入研究「TONIC: Towards oblivious neural inference compiler」主題。共同形成了獨特的指紋。