AI Paper Atlas

Research map / Adversarial robustness and security / Certified and empirical robustness

Perturbation-Induced Linearization: Constructing Unlearnable Data with Solely Linear Classifiers

Jinlin Liu, Wei Chen, Xiaojin Zhang

ICLR 2026 · Adversarial robustness and security › Certified and empirical robustness · approach: Decision-boundary geometry of robustness · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
pretrain the linear surrogate before generating perturbations (ablation)PIL · CIFAR-10 Rotation augmentationclean test accuracy: 27.99 -> 18.15Improved
pretrain the linear surrogate before generating perturbations (ablation)PIL · CIFAR-10 ChannelShuffle augmentationclean test accuracy: 34.13 -> 10.88Improved
pretrain the linear surrogate before generating perturbations (ablation)PIL · CIFAR-10 Cutout augmentationclean test accuracy: 10.90 -> 14.62Hurt
use PIL instead of CUDA* under budget-matched 8/255CUDA* · CIFAR-10 Basic augmentationclean test accuracy: 87.45 -> 12.87Improved
use linear surrogate (PIL) instead of deep surrogate REM for generation timeREM · CIFAR-10generation time (s): 54.46k -> 40.53Improved
increase perturbed fraction from 0.1 to 0.9PIL · CIFAR-10 partial perturbationclean test accuracy: 92.95 -> 76.97Hurt
train on 90% PIL-perturbed vs clean dataResNet-18 · CIFAR-10 FGSM 8/255accuracy drop: -62.44 -> -72.2Hurt
apply adversarial training against PIL at eps=2/255PIL · CIFAR-10clean test accuracy: 88.31 -> 78.15Hurt
use PIL instead of EM under JPEG50 compressionEM · CIFAR-10clean test accuracy: 55.12 -> 50.87Improved
add gradient orthogonality assumption: clean/perturbed gradients orthogonalPIL partial perturbation theory · CIFAR-10clean loss decreaseNo clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy