Toward Efficient Data-Free Unlearning
Abstract
Machine unlearning without access to real data distribution is challenging. The existing method based on data-free distillation achieved unlearning by filtering out synthetic samples containing forgetting information but struggled to distill the retaining-related knowledge efficiently. In this work, we analyze that such a problem is due to over-filtering, which reduces the synthesized retaining-related information. We propose a novel method, Inhibited Synthetic PostFilter (ISPF), to tackle this challenge from two perspectives: First, the Inhibited Synthetic, by reducing the synthesized forgetting information; Second, the PostFilter, by fully utilizing the retaining-related information in synthesized samples. Experimental results demonstrate that the proposed ISPF effectively tackles the challenge and outperforms existing methods.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
2 top-tier papers have cited it since. Know what they found before your reviewers do.
Lune can draw on
the full paper · the top-tier papers that cite it · related work
Free to start. No credit card required.
Lineage
- Fast Machine Unlearning without Retraining through Selective Synaptic Dampening cites Membership Inference Attacks Against Machine Learning Models
- Up to 100x Faster Data-Free Knowledge Distillation cites Data-Free Learning of Student Networks
- Synthetic Forgetting Without Access: A Few-Shot Zero-Glance Framework for Machine Unlearning cites SalUn: Empowering Machine Unlearning via Gradient-based Weight Saliency in Both Image Classification and Generation
- Efficient Source-free Unlearning via Energy-Guided Data Synthesis and Discrimination-Aware Multitask Optimization cites Fast Machine Unlearning without Retraining through Selective Synaptic Dampening
Cited by top-tier papers2
- Synthetic Forgetting Without Access: A Few-Shot Zero-Glance Framework for Machine UnlearningQipeng Song, Nan Yang, Ziqi Xu, Yue Li et al.AAAI 2026 · 2 citations
- Efficient Source-free Unlearning via Energy-Guided Data Synthesis and Discrimination-Aware Multitask OptimizationXiuyuan Wang, Chaochao Chen, Weiming Liu, Xinting Liao et al.ICML 2025
Builds on10
- Membership Inference Attacks Against Machine Learning ModelsReza Shokri, Marco Stronati, Congzheng Song, Vitaly ShmatikovS&P 2017 · 5,137 citations
- Data-Free Learning of Student NetworksHanting Chen, Yunhe Wang, Chang Xu, Zhaohui Yang et al.ICCV 2019 · 427 citations
- SalUn: Empowering Machine Unlearning via Gradient-based Weight Saliency in Both Image Classification and GenerationChongyu Fan, Jiancheng Liu, Yihua Zhang, Eric Wong et al.ICLR 2024 · 351 citations
- Fast Machine Unlearning without Retraining through Selective Synaptic DampeningJack Foster, Stefan Schoepf, Alexandra BrintrupAAAI 2024 · 208 citations
- Up to 100x Faster Data-Free Knowledge DistillationGongfan Fang, Kanya Mo, Xinchao Wang, Jie Song et al.AAAI 2022 · 103 citations
Related papers
- RUAGO: Effective and Practical Retain-Free Unlearning via Adversarial Attack and OOD GeneratorSangyong Lee, Sangjun Chung, Simon S. WooNeurIPS 2025 · 3 citations
- When Data-Free Knowledge Distillation Meets Non-Transferable Teacher: Escaping Out-of-Distribution Trap is All You NeedZiming Hong, Runnan Chen, Zengmao Wang, Bo Han et al.ICML 2025
- Always Be Dreaming: A New Approach for Data-Free Class-Incremental LearningJames Seale Smith, Yen-Chang Hsu, Jonathan C. Balloch, Yilin Shen et al.ICCV 2021 · 208 citations
- Remaining-data-free Machine Unlearning by Suppressing Sample ContributionXinwen Cheng, Zhehao Huang, Wenxing Zhou, Zhengbao He et al.ICLR 2026 · 11 citations
- Adversarial Mixup UnlearningZhuoyi Peng, Yixuan Tang, Yi YangICLR 2025
