MAS-Encryption and its Applications in Privacy-Preserving Classifiers

Homomorphic encryption (HE) schemes, such as fully homomorphic (FHE), support a number of useful computations on ciphertext in broad range applications, e-voting, private information retrieval, cloud security, and privacy protection. While FHE schemes do not require any interaction during computation, the key limitations are large expansion inefficiency. Thus, to overcome these limitations, we develop novel cryptographic tool, MAS-Encryption (MASE), real-value input secure computation multiply-add structure. The structures exist many important protocols, classifiers outsourced will explain how MASE can be used protect using two case study examples. Specifically, first example is privacy-preserving Naive Bayes classifier that achieve minimal risk, other vector machine. We prove constructed evaluate their performance real-world datasets. Experiments show our proposed scheme based efficient, sense an optimal tradeoff between efficiency communication interactions. avoid inefficiency paradigm.