Poster: Practical Privacy-Preserving Decision Tree Evaluation for Resource-Constrained Devices

We investigate the problem of private decision tree evaluation, which involves a server who holds a private decision tree, and a client who wants to classify its private attribute vector on the decision tree. The goal is to enable the client to learn the classification result while revealing nothing about both parties' inputs. We propose a novel secure two-party protocol for the problem of private decision tree evaluation based on symmetric encryption and oblivious transfer, which achieves higher efficiency and can be applied to scenarios involving resource-constrained devices.