Fully Homomorphic Encryption for Classification in Machine Learning
Seiko Arita, Shota Nakasato · 2017
Using fully homomorphic encryption scheme, we construct fully homomorphic encryption scheme FHE4GT that can homomorphically compute an encryption of the greater-than bit that indicates x > x' or not, given two ciphertexts c and c' of x and x', respectively, without knowing the secret key. Then, we construct homomorphic classifier homClassify that can homomorphically classify a given encrypted data without decrypting it, using machine learned parameters.