Ferrari: Federated Feature Unlearning via Optimizing Feature Sensitivity

TL;DR

This paper introduces Ferrari, a federated learning framework that unlearns specific features by minimizing their sensitivity, ensuring privacy and model integrity without requiring all clients' participation.

Contribution

Ferrari is the first federated feature unlearning method that minimizes feature sensitivity based on Lipschitz continuity, addressing limitations of influence function-based approaches.

Findings

Ferrari effectively unlearns sensitive, backdoor, and biased features.

Experimental results show Ferrari outperforms existing methods in unlearning effectiveness.

Theoretical analysis confirms the robustness of the feature sensitivity minimization.

Abstract

The advent of Federated Learning (FL) highlights the practical necessity for the right to be forgotten for all clients, allowing them to request data deletion from the machine learning models service provider. This necessity has spurred a growing demand for Federated Unlearning (FU). Feature unlearning has gained considerable attention due to its applications in unlearning sensitive, backdoor, and biased features. Existing methods employ the influence function to achieve feature unlearning, which is impractical for FL as it necessitates the participation of other clients, if not all, in the unlearning process. Furthermore, current research lacks an evaluation of the effectiveness of feature unlearning. To address these limitations, we define feature sensitivity in evaluating feature unlearning according to Lipschitz continuity. This metric characterizes the model outputs rate of change or sensitivity to perturbations in the input feature. We then propose an effective federated feature unlearning framework called Ferrari, which minimizes feature sensitivity. Extensive experimental results and theoretical analysis demonstrate the effectiveness of Ferrari across various feature unlearning scenarios, including sensitive, backdoor, and biased features. The code is publicly available at https://github.com/OngWinKent/Federated-Feature-Unlearning