New Approaches to Federated XGBoost Learning for Privacy-Preserving Data Analysis

In this paper, we propose a new privacy-preserving machine learning algorithm called Federated-Learning XGBoost (FL-XGBoost), in which a federated learning scheme is introduced into XGBoost, a state-of-the-art gradient boosting decision tree model. The proposed FL-XGBoost can train a sensitive task to be solved among different entities without revealing their own data. The proposed FL-XGBoost can achieve significant reduction in the number of communications between entities by exchanging decision tree models. In our experiments, we carry out the performance comparison between FL-XGBoost and a different federated learning approach to XGBoost called FATE. The experimental results show that the proposed method can achieve high prediction accuracy with less communication even if the number of entities is increase.