Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2602.01150 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866918486576463872 |
|---|---|
| author | Sun, Jialong Wei, Zeming Zou, Jiaxuan Gong, Jiacheng Fu, Jie Dong, Chengyang Xu, Heng Li, Jialong Liu, Bo |
| author_facet | Sun, Jialong Wei, Zeming Zou, Jiaxuan Gong, Jiacheng Fu, Jie Dong, Chengyang Xu, Heng Li, Jialong Liu, Bo |
| contents | Machine unlearning (MU) is essential for enforcing the right to be forgotten in machine learning systems. A key challenge of MU is how to reliably audit whether a model has truly forgotten specified training data. Membership Inference Attacks (MIAs) are widely used for unlearned model auditing, where samples that evade membership detection are regarded as successfully forgotten. We show this assumption is fundamentally flawed: failed membership inference does not imply true forgetting. We prove that unlearned samples occupy fundamentally different positions in the feature space than non-member samples, making this alignment bias unavoidable and unobservable, which leads to systematically optimistic evaluations of unlearning performance. Meanwhile, training shadow models for MIA incurs substantial computational overhead. To address both limitations, we propose Statistical Membership Inference (SMI), a training-free auditing framework that reformulates auditing as estimating the non-member mixture proportion in the unlearned feature distribution. Beyond estimating the forgetting rate, SMI also provides bootstrap reference ranges for quantified auditing reliability. Extensive experiments show that SMI consistently outperforms all MIA-based baselines, with no shadow model training required. Overall, SMI establishes a principled and efficient alternative to MIA-based auditing methods, with both theoretical guarantees and strong empirical performance. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_01150 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | SMI: Statistical Membership Inference for Reliable Unlearned Model Auditing Sun, Jialong Wei, Zeming Zou, Jiaxuan Gong, Jiacheng Fu, Jie Dong, Chengyang Xu, Heng Li, Jialong Liu, Bo Machine Learning Artificial Intelligence Cryptography and Security Computer Vision and Pattern Recognition Optimization and Control Machine unlearning (MU) is essential for enforcing the right to be forgotten in machine learning systems. A key challenge of MU is how to reliably audit whether a model has truly forgotten specified training data. Membership Inference Attacks (MIAs) are widely used for unlearned model auditing, where samples that evade membership detection are regarded as successfully forgotten. We show this assumption is fundamentally flawed: failed membership inference does not imply true forgetting. We prove that unlearned samples occupy fundamentally different positions in the feature space than non-member samples, making this alignment bias unavoidable and unobservable, which leads to systematically optimistic evaluations of unlearning performance. Meanwhile, training shadow models for MIA incurs substantial computational overhead. To address both limitations, we propose Statistical Membership Inference (SMI), a training-free auditing framework that reformulates auditing as estimating the non-member mixture proportion in the unlearned feature distribution. Beyond estimating the forgetting rate, SMI also provides bootstrap reference ranges for quantified auditing reliability. Extensive experiments show that SMI consistently outperforms all MIA-based baselines, with no shadow model training required. Overall, SMI establishes a principled and efficient alternative to MIA-based auditing methods, with both theoretical guarantees and strong empirical performance. |
| title | SMI: Statistical Membership Inference for Reliable Unlearned Model Auditing |
| topic | Machine Learning Artificial Intelligence Cryptography and Security Computer Vision and Pattern Recognition Optimization and Control |
| url | https://arxiv.org/abs/2602.01150 |