AI Paper Atlas

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

NAPPure: Adversarial Purification for Robust Image Classification under Non-Additive Perturbations

Junjie Nan, Jianing Li, Wei Chen, Mingkun Zhang, Xueqi Cheng

ICCV 2025 · Adversarial robustness and security › Certified and empirical robustness · approach: Augmentation and diffusion purification · 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
use NAPPure instead of DiffPureGTSRB-CNN · GTSRB average over attacksrobust accuracy (%): 47.61 -> 73.93Improved
use NAPPure instead of DiffPureWideResNet · CIFAR-10 average over attacksrobust accuracy (%): 57.82 -> 68.55Improved
use NAPPure instead of DiffPure on large-scale dataResNet-50 · ImageNet average over attacks (512 images)robust accuracy (%): 29.62 -> 37.79Improved
use NAPPure instead of adversarial training under additive attackGTSRB-CNN · GTSRB additive attackrobust accuracy (%): 47.85 -> 83.20Improved
use NAPPure instead of LM under additive attackGTSRB-CNN · GTSRB additive attackrobust accuracy (%): 79.07 -> 83.20No clear effect
use type-agnostic NAPPure-joint instead of type-specific NAPPureNAPPure · GTSRB average over attacksrobust accuracy (%): 73.93 -> 59.29Hurt
use NAPPure-joint interpolation instead of direct composition under composite attack (ablation)NAPPure · GTSRB composite attack of all 4 perturbation typesrobust accuracy (%): 37.10 -> 54.49Improved
increase purification iterations from 100 to 500 (ablation)NAPPure · GTSRB patch attackrobust accuracy (%): 63.48% -> 74.22%Improved
extend purification iterations from 500 to 1000 (ablation)NAPPure · GTSRB patch attackrobust accuracy (%): 74.22% -> 60.74%Hurt
replace 3-layer CNN auxiliary model with ResNet-18 (ablation)NAPPure · GTSRB patch attackrobust accuracy (%): 74.22% -> 71.29%No clear effect
enlarge attack patch size from 5x5 to 9x9 against fixed defense (ablation)NAPPure · GTSRB patch attackrobust accuracy (%): 85.16% -> 67.97%Hurt
enlarge blur attack kernel from 3x3 to 5x5 against 5x5 defense kernel (ablation)NAPPure · GTSRB blur attackrobust accuracy (%): 91.80% -> 86.91%Hurt

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