A machine learning model based on heterogeneous data

Abstract

Big data is widely used in many areas of business. The information between organizations is systematically reproduced and processed by data, and the collected data differs significantly in attributes. By composing heterogeneous data sets, they complement each other, therefore, data exchange between organizations is necessary. In a machine learning collaborative learning process based on heterogeneous data, the current schema has many challenges, including efficiency, security, and availability in real-world situations. In this paper, we propose a secure SVM learning mechanism based on the consortium blockchain and a threshold homomorphic encryption algorithm. By implementing the consortium’s blockchain, it is possible to build a decentralized data exchange platform, and also to develop a secure algorithm for the support-vector machine classifier based on threshold homomorphic encryption