A preprint paper is available on Arxiv

A paper entitled CHEETAH: An Ultra-Fast, Approximation-Free, and Privacy-Preserved Neural Network Framework based on Joint Obscure Linear and Nonlinear Computations is available on arxiv. This paper is about the way to accelerate the inference calculation with privacy concerns in Machine Leaning as a Service (MLaaS). The basic idea is to combine the calculation of linear and non-linear functions in neural network, which ignores the traditional propagation from exact linear result to non-linear function. We show that this joint computation for linear and non-linear functions largely improves the efficiency for privacy-preserved MLaaS.