BootNet: Homomorphic CNN Inference with Convolution and ReLU Fused in Bootstrapping

Fully homomorphic encryption (FHE) enables privacy-preserving neural network inference but suffers from high overhead from homomorphic convolutions, polynomial activation approximations, and CKKS bootstrapping. This paper presents BootNet, a unified framework that fuses all three operations into a single bootstrapping invocation per CNN layer, achieving convolution, ReLU, and noise refresh simultaneously.

Prior works are able to fuse convolution into bootstrapping using CinS encoding (NeuJeans,