This repository provides a basic example of how to compute the decision function of Support Vector Machines (SVMs) on encrypted input samples using the CKKS scheme as implemented in OpenFHE. The SVMs are built with linear and polynomial kernels. Note that only the decision function is computed homomorphically, the client will need to evaluate the sign function (for classification tasks) after receiving the encrypted decision function result from the server. For regression tasks, the client does not have to compute anything after receiving the encrypted result.

3 Likes

A Python version of homomorphic SVM inference is available at GitHub - openfheorg/python-svm-examples Contributed by @reneroliveira