Saved in:
Bibliographic Details
Main Authors: Liu, Jinlin, Chen, Wei, Zhang, Xiaojin
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2601.19967
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910004831846400
author Liu, Jinlin
Chen, Wei
Zhang, Xiaojin
author_facet Liu, Jinlin
Chen, Wei
Zhang, Xiaojin
contents Collecting web data to train deep models has become increasingly common, raising concerns about unauthorized data usage. To mitigate this issue, unlearnable examples introduce imperceptible perturbations into data, preventing models from learning effectively. However, existing methods typically rely on deep neural networks as surrogate models for perturbation generation, resulting in significant computational costs. In this work, we propose Perturbation-Induced Linearization (PIL), a computationally efficient yet effective method that generates perturbations using only linear surrogate models. PIL achieves comparable or better performance than existing surrogate-based methods while reducing computational time dramatically. We further reveal a key mechanism underlying unlearnable examples: inducing linearization to deep models, which explains why PIL can achieve competitive results in a very short time. Beyond this, we provide an analysis about the property of unlearnable examples under percentage-based partial perturbation. Our work not only provides a practical approach for data protection but also offers insights into what makes unlearnable examples effective.
format Preprint
id arxiv_https___arxiv_org_abs_2601_19967
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Perturbation-Induced Linearization: Constructing Unlearnable Data with Solely Linear Classifiers
Liu, Jinlin
Chen, Wei
Zhang, Xiaojin
Machine Learning
Artificial Intelligence
Collecting web data to train deep models has become increasingly common, raising concerns about unauthorized data usage. To mitigate this issue, unlearnable examples introduce imperceptible perturbations into data, preventing models from learning effectively. However, existing methods typically rely on deep neural networks as surrogate models for perturbation generation, resulting in significant computational costs. In this work, we propose Perturbation-Induced Linearization (PIL), a computationally efficient yet effective method that generates perturbations using only linear surrogate models. PIL achieves comparable or better performance than existing surrogate-based methods while reducing computational time dramatically. We further reveal a key mechanism underlying unlearnable examples: inducing linearization to deep models, which explains why PIL can achieve competitive results in a very short time. Beyond this, we provide an analysis about the property of unlearnable examples under percentage-based partial perturbation. Our work not only provides a practical approach for data protection but also offers insights into what makes unlearnable examples effective.
title Perturbation-Induced Linearization: Constructing Unlearnable Data with Solely Linear Classifiers
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2601.19967