Jellyfish: Zero-Shot Federated Unlearning Scheme with Knowledge Disentanglement

TL;DR

Jellyfish introduces a zero-shot federated unlearning method that uses synthetic data, knowledge disentanglement, and a comprehensive loss function to effectively forget specific data while preserving model utility.

Contribution

It presents a novel zero-shot federated unlearning framework with a unique knowledge disentanglement and model repair mechanism, enhancing privacy and utility.

Findings

Effective forgetting of data with minimal utility loss.

Robustness validated across diverse experimental settings.

Synthetic proxy data successfully replaces access to original data.

Abstract

With the increasing importance of data privacy and security, federated unlearning emerges as a new research field dedicated to ensuring that once specific data is deleted, federated learning models no longer retain or disclose related information. In this paper, we propose a zero-shot federated unlearning scheme, named Jellyfish. It distinguishes itself from conventional federated unlearning frameworks in four key aspects: synthetic data generation, knowledge disentanglement, loss function design, and model repair. To preserve the privacy of forgotten data, we design a zero-shot unlearning mechanism that generates error-minimization noise as proxy data for the data to be forgotten. To maintain model utility, we first propose a knowledge disentanglement mechanism that regularises the output of the final convolutional layer by restricting the number of activated channels for the data to be forgotten and encouraging activation sparsity. Next, we construct a comprehensive loss function that incorporates multiple components, including hard loss, confusion loss, distillation loss, model weight drift loss, gradient harmonization, and gradient masking, to effectively align the learning trajectories of the objectives of ``forgetting" and ``retaining". Finally, we propose a zero-shot repair mechanism that leverages proxy data to restore model accuracy within acceptable bounds without accessing users' local data. To evaluate the performance of the proposed zero-shot federated unlearning scheme, we conducted comprehensive experiments across diverse settings. The results validate the effectiveness and robustness of the scheme.