Penetralium: Privacy-preserving and memory-efficient neural network inference at the edge

Mengda Yang, Wenzhe Yi, Juan Wang, Hongxin Hu, Xiaoyang Xu, Ziang Li

Abstract: Highlights•A system-algorithm jointly optimized model inference system at the edge.•Building computational pipeline that can be adapted to underlying trusted components.•Lightweight perturbation policy to protect against advanced privacy attacks.•The system provides strong security guarantees with reasonable performance.