Training Support Vector Machines with privacy-protected data

• Published in: Pattern recognition Vol. 72; pp. 93 - 107
• Main Authors: González-Serrano, Francisco-Javier, Navia-Vázquez, Ángel, Amor-Martín, Adrián
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.12.2017
• Subjects: Homomorphic encryption; Machine learning; Privacy protection; Support Vector Machines; Privacy protection; Homomorphic encryption; Support Vector Machines; Machine learning
• ISSN: 0031-3203; 1873-5142
• DOI: 10.1016/j.patcog.2017.06.016

•Multiple-key encrypted machine learning scenario.•Standard authorization protocol (OAuth 2.0) to get access to encrypted data.•A minimal set of outsourced operations to optimize the encryption/decryption hardware (CryptoProcessor).•Semiparametric SVM scheme that avoids the use of private instances as part of the model.•Analysis of the SVMs performance under thenite-precission conditions required by cryptosystems. In this paper, we address a machine learning task using encrypted training data. Our basic scenario has three parties: Data Owners, who own private data; an Application, which wants to train and use an arbitrary machine learning model on the Users’ data; and an Authorization Server, which provides Data Owners with public and secret keys of a partial homomorphic cryptosystem (that protects the privacy of their data), authorizes the Application to get access to the encrypted data, and assists it in those computations not supported by the partial homomorphism. As machine learning model, we have selected the Support Vector Machine (SVM) due to its excellent performance in supervised classification tasks. We evaluate two well known SVM algorithms, and we also propose a new semiparametric SVM scheme better suited for the privacy-protected scenario. At the end of the paper, a performance analysis regarding the accuracy and the complexity of the developed algorithms and protocols is presented.