Outsourced Secure ID3 Decision Tree Algorithm over Horizontally Partitioned Datasets with Consortium Blockchain

Abstract

Due to the capacity of storing massive data and providing huge computing resources, cloud computing has been a desirable platform to assist machine learning in multiple-data-owners scenarios. However, the issue of data privacy is far from being well solved and thus has been a general concern in the cloud-assisted machine learning. For example, in the existing cloud-assisted decision tree classification algorithms, it is very hard to guarantee data privacy since all data owners have to aggregate their data to the cloud platform for model training. In this paper, we investigate the possibility of training a decision tree in the scenario that the distributed data are stored locally in each data owner, where the privacy of the original data can be guaranteed in a more intuitive approach. Specifically, we present a positive answer to the above issue by presenting a privacy-preserving ID3 training scheme using Gini index over horizontally partitioned datasets by multiple data owners. Since each data owner cannot directly divide the local dataset according to the best attributes selected, a consortium blockchain and a homomorphic encryption algorithm are employed to ensure the privacy and usability of the distributed data. Security analysis indicates that our scheme can preserve the privacy of the original data and the intermediate values. Moreover, extensive experiments show that our scheme can achieve the same result compared with the original ID3 decision tree algorithm while additionally preserving data privacy, and calculation time overhead and communication time overhead on data owners decrease greatly.