Funshade: Function Secret Sharing for Two-Party Secure Thresholded Distance Evaluation

We propose a novel privacy-preserving, two-party computation of various distance metrics (e.g., Hamming distance, Scalar Product) followed by comparison with fixed threshold, which is known as one the most useful and popular building blocks for many different applications including machine learning, biometric matching, etc. Our solution builds upon recent advances in function secret sharing makes use an optimized version arithmetic sharing. Thanks to this combination, our new named Funshade first require only round communication two ring elements online phase, outperforming all prior state-of-the-art schemes while relying on lightweight cryptographic primitives. Lastly, we implement from scratch portable C expose it Python, testifying its high performance running secure identification against database 1 million records ~10 seconds full correctness 32-bit precision, without parallelization.