SecDT: Privacy-Preserving Outsourced Decision Tree Classification Without Polynomial Forms in Edge-Cloud Computing

In the era of cloud computing with security, how to outsource evaluation services to a cloud server but preserve the model privacy is an important issue. In this paper, we study how to perform decision tree evaluation on the cloud server while achieving privacy preservation with support by edges. Existing research mainly focuses on treating the decision tree model as a polynomial form and using homomorphic encryption to ensure security, and both of them are reckoned with computational overhead. However, due to the environment that devices are getting more but smaller while the network is getting faster (like IoT or 5G), there should be some more suitable solutions to solve this problem. Therefore, we aim for constructing an outsourced decision tree classification with lightweight cryptographic tools to preserve data privacy. The main technique is to build secret sharing-based protocols for the model owner, service user, and cloud server. We trade off the communication rounds with the computational cost to reduce the overhead of the cloud server and users. We conduct some experiments with real-world datasets, which show that our scheme has desirable utility and efficiency.