Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2412.00126 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910866320916480 |
|---|---|
| author | Zhou, Lei Zhu, Youwen Xue, Qiao Zhang, Ji Zhang, Pengfei |
| author_facet | Zhou, Lei Zhu, Youwen Xue, Qiao Zhang, Ji Zhang, Pengfei |
| contents | Recently, the enactment of ``right to be forgotten" laws and regulations has imposed new privacy requirements on federated learning (FL). Researchers aim to remove the influence of certain data from the trained model without training from scratch through federated unlearning (FU). While current FU research has shown progress in enhancing unlearning efficiency, it often results in degraded model performance upon achieving the goal of data unlearning, necessitating additional steps to recover the performance of the unlearned model. Moreover, these approaches also suffer from many shortcomings such as high consumption of computational and storage resources. To this end, we propose a streamlined federated unlearning approach (SFU) aimed at effectively removing the influence of the target data while preserving the model performance on the retained data without degradation. We design a practical multi-teacher system that achieves both target data influence removal and model performance preservation by guiding the unlearned model through several distinct teacher models. SFU is both computationally and storage-efficient, highly flexible, and generalizable. We conduct extensive experiments on both image and text benchmark datasets. The results demonstrate that SFU significantly improves time and communication efficiency compared to the benchmark retraining method and significantly outperforms existing SOTA methods. Additionally, we verify the effectiveness of SFU using the backdoor attack. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2412_00126 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Streamlined Federated Unlearning: Unite as One to Be Highly Efficient Zhou, Lei Zhu, Youwen Xue, Qiao Zhang, Ji Zhang, Pengfei Machine Learning Recently, the enactment of ``right to be forgotten" laws and regulations has imposed new privacy requirements on federated learning (FL). Researchers aim to remove the influence of certain data from the trained model without training from scratch through federated unlearning (FU). While current FU research has shown progress in enhancing unlearning efficiency, it often results in degraded model performance upon achieving the goal of data unlearning, necessitating additional steps to recover the performance of the unlearned model. Moreover, these approaches also suffer from many shortcomings such as high consumption of computational and storage resources. To this end, we propose a streamlined federated unlearning approach (SFU) aimed at effectively removing the influence of the target data while preserving the model performance on the retained data without degradation. We design a practical multi-teacher system that achieves both target data influence removal and model performance preservation by guiding the unlearned model through several distinct teacher models. SFU is both computationally and storage-efficient, highly flexible, and generalizable. We conduct extensive experiments on both image and text benchmark datasets. The results demonstrate that SFU significantly improves time and communication efficiency compared to the benchmark retraining method and significantly outperforms existing SOTA methods. Additionally, we verify the effectiveness of SFU using the backdoor attack. |
| title | Streamlined Federated Unlearning: Unite as One to Be Highly Efficient |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2412.00126 |