1 paper
Huaming Ling, Ying Wang, Si Chen +1
We address two fundamental challenges in adapting general deep CNNs for FHE-based inference: approximating non-linear activations such as ReLU with low-degree polynomials while min…