Increasing trust in AI through privacy preservation and model explainability: Federated Learning of Fuzzy Regression Trees

José Luis Corcuera Bárcena, Pietro Ducange, Francesco Marcelloni, Alessandro Renda

Abstract: Highlights•A novel approach for Federated Learning of Fuzzy Regression Trees is introduced.•The approach fulfills two requirements for trustworthy AI: privacy and transparency.•Federated FRT outperforms FRTs learned locally on each participant.•Federated FRT achieves substantially equivalent results compared to centralized FRT.•The approach is adopted in the context of an innovative automotive case-study.