Skip to content

Do Influence-Derived Data Perturbations Enable Machine Unlearning? A Controlled Study of Three Plausible Roles

Sep 2026 · 0 citations · 38 references
Computer Science

TL;DR

Deep Perturbation Learning is evaluated in three roles in which prior work has positioned it for machine unlearning: a direct deletion signal (the strongest claim), a utility-preserving regularizer, and a warm start for adversarial unlearning.

Abstract

We evaluate Deep Perturbation Learning (DPL), which perturbs training images and labels along influence-derived directions, in three roles in which prior work has positioned it for machine unlearning: a direct deletion signal (the strongest claim), a utility-preserving regularizer, and a warm start for adversarial unlearning. Evidence for the weaker roles has been used to support the stronger one, so we test each role separately under a matched protocol with exact-seed retraining baselines. An audit of the public implementation identifies two correctness issues: image directions are computed on augmented, normalized tensors but applied to raw images, and the label perturbation falls below float32 resolution, leaving labels unchanged. After correcting the image-perturbation pipeline, DPL fails the direct-deletion criterion on CIFAR-10/ResNet-18 in all three paired seeds. Its utility effects are inconsistent in sign across seeds, and once direction-computation time is counted it underperforms simple warm-start baselines. A one-seed Tiny ImageNet check likewise does not favor DPL as a regularizer or warm start; preprocessing inconsistencies in the released code make the direct comparison there inconclusive. These results cover random instance deletion only and do not rule out influence-based methods in other deletion regimes. We release a role-matched evaluation protocol and an audit checklist for perturbation-based deletion claims.

View source

Similar papers

#machine learning Preprint Sep 2026

The BatchNorm Illusion: Diagnosing Normalization Artifacts in Machine Unlearning Evaluation

A previously undocumented confound in how unlearning is evaluated on BatchNorm-based architectures is identified: a single forward pass over retain data, an operation that modifies no weight, can deterministically rewrite the model's normalization state and reverse the apparent surface-metric forgetting.

Aaryaman Kalani, Murari Mandal, Dhruv Kumar et al. · 1 citation
2026

DeepU: Deeper Granular Within-Layer Machine Unlearning

Machine unlearning (MU) aims to remove the influence of selected data from trained models, offering an efficient alternative to full retraining. With the rise of increasingly stringent privacy regulations, including the right to be forgotten, machine learning models must incorporate mechanisms that ensure compliance wh...

Anudeep Vurity, Zhi-Sheng Yan, Massimiliano Albanese · 0 citations
Preprint Aug 2026

Can LLMs Truly Forget? Revealing Unlearning Gaps Through Adversarial Evaluation

Strong standard-metric performance alone is insufficient to establish robustness after unlearning and motivate adversarial stress-testing as a complementary component of unlearning evaluation, showing a substantial gap between clean-query forgetting and adversarial robustness.

A. Gupta, Hima Varshini Surisetty, Sreevidya Bollineni et al. · 1 citation
Open access Sep 2026

DWAT: Density-Weighted Adversarial Training for Robustness Beyond the Training Perturbation Budget

Density-Weighted Adversarial Training is presented, a plug-in framework that spreads training over a set of sampled ϵ-balls and reweights each candidate adversarial example by a Gaussian density of its distance from the benign sample, and it is proved that the population objective upper-bounds the worst-case adversaria...

Jie-Ying Huang, Rui-Ming Zhu, Jia Xu et al. · 0 citations
Preprint Aug 2026

Cross-Domain Generalization in Machine Unlearning via Label-Conditioned Energy Magnitude Regularization

This paper studies what happens to the rest of the model when a class is forgotten, using a label-conditioned energy-based model (EBM) that assigns per-class energies, making the effect directly observable.

Syed Ali Ahmed, Syed Bilal Ahsan, Muhammad Zaigham Zaheer National University of Computer et al. · 0 citations
#artificial intelligence Preprint Aug 2026

Forget or Fine-tune? A Comparative Study of Machine Unlearning Strategies for Noisy Label Correction

This work conducts a comparative empirical study of five MU methods across symmetric, asymmetric, instance-dependent, and open-set noise on CIFAR-10, CIFAR-100, and the real-world noisy dataset Food-101N and finds that the appropriate unlearning strategy is conditioned on the noise structure.

J. L. Sant'Ana, Filipe R. Cordeiro · 0 citations

Related blog posts

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.