Towards Explainable Federated Learning: Understanding the Impact of Differential Privacy

TL;DR

This paper tackles the challenge of achieving both data privacy and explainability in Federated Learning by proposing Federated EXplainable Trees with Differential Privacy (FEXT-DP), a bagging-based, DT‑driven FL approach that injects Differential Privacy via the exponential mechanism in split selection. The server aggregates only trees meeting a minimum accuracy threshold, while clients iteratively refine the ensemble across rounds. Empirical results on the AEPD dataset show that FEXT-DP can deliver faster convergence and competitive $MSE$ relative to FedAVG, with DP introducing measurable impacts on interpretability as shown by $MDI$ analyses. The work highlights a practical privacy–explainability trade-off and outlines clear directions for improving efficiency and explainability in future iterations.

Abstract

Data privacy and eXplainable Artificial Intelligence (XAI) are two important aspects for modern Machine Learning systems. To enhance data privacy, recent machine learning models have been designed as a Federated Learning (FL) system. On top of that, additional privacy layers can be added, via Differential Privacy (DP). On the other hand, to improve explainability, ML must consider more interpretable approaches with reduced number of features and less complex internal architecture. In this context, this paper aims to achieve a machine learning (ML) model that combines enhanced data privacy with explainability. So, we propose a FL solution, called Federated EXplainable Trees with Differential Privacy (FEXT-DP), that: (i) is based on Decision Trees, since they are lightweight and have superior explainability than neural networks-based FL systems; (ii) provides additional layer of data privacy protection applying Differential Privacy (DP) to the Tree-Based model. However, there is a side effect adding DP: it harms the explainability of the system. So, this paper also presents the impact of DP protection on the explainability of the ML model. The carried out performance assessment shows improvements of FEXT-DP in terms of a faster training, i.e., numbers of rounds, Mean Squared Error and explainability.

Figures (5)

• Figure 1: Overview.
• Figure 2: Diagram: Detailed interaction between FL server and clients.
• Figure 3: The impact of Differential Privacy on Performance Over the Training Rounds.
• Figure 4: The impact of Differential Privacy on Explainability Based on the Mean Decrease in Impurity (MDI).
• Figure 5: Explainability Analysis of FEXT-DP: Heatmap of Mean Decrease in Impurity (MDI).