Efficient Source-free Unlearning via Energy-Guided Data Synthesis and Discrimination-Aware Multitask Optimization

Xiuyuan Wang, Chaochao Chen, Weiming Liu, Xinting Liao, Fan Wang, Xiaolin Zheng
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:62518-62528, 2025.

Abstract

With growing privacy concerns and the enforcement of data protection regulations, machine unlearning has emerged as a promising approach for removing the influence of forget data while maintaining model performance on retain data. However, most existing unlearning methods require access to the original training data, which is often impractical due to privacy policies, storage constraints, and other limitations. This gives rise to the challenging task of source-free unlearning, where unlearning must be accomplished without accessing the original training data. Few existing source-free unlearning methods rely on knowledge distillation and model retraining, which impose substantial computational costs. In this work, we propose the Data Synthesis-based Discrimination-Aware (DSDA) unlearning framework, which enables efficient source-free unlearning in two stages: (1) Accelerated Energy-Guided Data Synthesis (AEGDS), which employs Langevin dynamics to model the training data distribution while integrating Runge–Kutta methods and momentum to enhance efficiency. (2) Discrimination-Aware Multitask Optimization (DAMO), which refines the feature distribution of retain data and mitigates the gradient conflicts among multiple unlearning objectives. Extensive experiments on three benchmark datasets demonstrate that DSDA outperforms existing unlearning methods, validating its effectiveness and efficiency in source-free unlearning.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-wang25m, title = {Efficient Source-free Unlearning via Energy-Guided Data Synthesis and Discrimination-Aware Multitask Optimization}, author = {Wang, Xiuyuan and Chen, Chaochao and Liu, Weiming and Liao, Xinting and Wang, Fan and Zheng, Xiaolin}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {62518--62528}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/wang25m/wang25m.pdf}, url = {https://proceedings.mlr.press/v267/wang25m.html}, abstract = {With growing privacy concerns and the enforcement of data protection regulations, machine unlearning has emerged as a promising approach for removing the influence of forget data while maintaining model performance on retain data. However, most existing unlearning methods require access to the original training data, which is often impractical due to privacy policies, storage constraints, and other limitations. This gives rise to the challenging task of source-free unlearning, where unlearning must be accomplished without accessing the original training data. Few existing source-free unlearning methods rely on knowledge distillation and model retraining, which impose substantial computational costs. In this work, we propose the Data Synthesis-based Discrimination-Aware (DSDA) unlearning framework, which enables efficient source-free unlearning in two stages: (1) Accelerated Energy-Guided Data Synthesis (AEGDS), which employs Langevin dynamics to model the training data distribution while integrating Runge–Kutta methods and momentum to enhance efficiency. (2) Discrimination-Aware Multitask Optimization (DAMO), which refines the feature distribution of retain data and mitigates the gradient conflicts among multiple unlearning objectives. Extensive experiments on three benchmark datasets demonstrate that DSDA outperforms existing unlearning methods, validating its effectiveness and efficiency in source-free unlearning.} }
Endnote
%0 Conference Paper %T Efficient Source-free Unlearning via Energy-Guided Data Synthesis and Discrimination-Aware Multitask Optimization %A Xiuyuan Wang %A Chaochao Chen %A Weiming Liu %A Xinting Liao %A Fan Wang %A Xiaolin Zheng %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-wang25m %I PMLR %P 62518--62528 %U https://proceedings.mlr.press/v267/wang25m.html %V 267 %X With growing privacy concerns and the enforcement of data protection regulations, machine unlearning has emerged as a promising approach for removing the influence of forget data while maintaining model performance on retain data. However, most existing unlearning methods require access to the original training data, which is often impractical due to privacy policies, storage constraints, and other limitations. This gives rise to the challenging task of source-free unlearning, where unlearning must be accomplished without accessing the original training data. Few existing source-free unlearning methods rely on knowledge distillation and model retraining, which impose substantial computational costs. In this work, we propose the Data Synthesis-based Discrimination-Aware (DSDA) unlearning framework, which enables efficient source-free unlearning in two stages: (1) Accelerated Energy-Guided Data Synthesis (AEGDS), which employs Langevin dynamics to model the training data distribution while integrating Runge–Kutta methods and momentum to enhance efficiency. (2) Discrimination-Aware Multitask Optimization (DAMO), which refines the feature distribution of retain data and mitigates the gradient conflicts among multiple unlearning objectives. Extensive experiments on three benchmark datasets demonstrate that DSDA outperforms existing unlearning methods, validating its effectiveness and efficiency in source-free unlearning.
APA
Wang, X., Chen, C., Liu, W., Liao, X., Wang, F. & Zheng, X.. (2025). Efficient Source-free Unlearning via Energy-Guided Data Synthesis and Discrimination-Aware Multitask Optimization. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:62518-62528 Available from https://proceedings.mlr.press/v267/wang25m.html.

Related Material