Ferrari: Federated Feature Unlearning via Optimizing Feature Sensitivity
This addresses the need for efficient data deletion in federated learning, enabling clients to request removal of specific features without involving others, though it is incremental as it builds on existing unlearning concepts.
The paper tackles the problem of feature unlearning in federated learning by proposing Ferrari, a framework that minimizes feature sensitivity to achieve unlearning without requiring participation from other clients, and demonstrates its effectiveness across various scenarios including sensitive, backdoor, and biased features.
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