Secure and Non-interactive k-NN Classifier Using Symmetric Fully Homomorphic Encryption

“Machine learning as a service” (MLaaS) in the cloud accelerates adoption of machine techniques. Nevertheless, externalization data on raises serious vulnerability issue because it requires disclosing private to provider. This paper deals with this problem and brings solution for K-nearest neighbors (k-NN) algorithm homomorphic encryption scheme (called TFHE) by operating end-to-end encrypted while preserving privacy. The proposed addresses all stages k-NN fully data, including majority vote class-label assignment. Unlike existing techniques, our does not require intermediate interactions between server client when executing classification task. Our has been assessed quantitative variables demonstrated its efficiency large relevant real-world sets scaling well across different parameters simulated data.