# Adversarial Defense on ImageNet

Source: [Adversarial Defense on ImageNet](<https://www.sota2.com/research/sota/adversarial-defense-on-imagenet>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-02-26T05:27:51.369Z

Evaluation of model performance under adversarial defense scenarios on the ImageNet dataset, measured by accuracy.

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Adversarial Defense |
| Dataset (including any stated split) | ImageNet |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 5 of 5 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Accuracy](<https://www.sota2.com/research/sota/adversarial-defense-on-imagenet.md?metric=Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| CutMix + MoEx | 41 | Not recorded | 68fa699a1e209131cea23030 |
| Shape-Texture Debiased Training | 27.4 | Not recorded | 68fa699a1e209131cea23030 |
| DeepAugment + AugMix | 18.8 | Not recorded | 68fa699a1e209131cea23030 |
| ResNet-50 | 17.1 | Not recorded | 68fa699a1e209131cea23030 |
| SIN | 5.6 | Not recorded | 68fa699a1e209131cea23030 |

### CutMix + MoEx

Accuracy: 41. Result ID: 698dc2c88b517eb433239079.

[Reporting paper](<https://www.sota2.com/research/paper/68fa699a1e209131cea23030.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa699a1e209131cea2301f.md>)

[Implementation](<https://github.com/Boyiliee/MoEx>)

Recorded evaluation configuration: {"Backbone":"ResNet-50"}. This record does not assert independent reproduction.

### Shape-Texture Debiased Training

Accuracy: 27.4. Result ID: 698dc2c88b517eb43323907c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa699a1e209131cea23030.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa699a1e209131cea23030.md>)

[Implementation](<https://github.com/LiYingwei/ShapeTextureDebiasedTraining>)

Recorded evaluation configuration: {"Backbone":"ResNet-50"}. This record does not assert independent reproduction.

### DeepAugment + AugMix

Accuracy: 18.8. Result ID: 698dc2c88b517eb43323907a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa699a1e209131cea23030.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6a751e209131cea28405.md>)

[Implementation](<https://github.com/hendrycks/imagenet-r>)

Recorded evaluation configuration: {"Backbone":"ResNet-50"}. This record does not assert independent reproduction.

### ResNet-50

Accuracy: 17.1. Result ID: 698dc2c88b517eb433239078.

[Reporting paper](<https://www.sota2.com/research/paper/68fa699a1e209131cea23030.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa62321e209131ce9f8328.md>)

Recorded evaluation configuration: {"Backbone":"ResNet-50"}. This record does not assert independent reproduction.

### SIN

Accuracy: 5.6. Result ID: 698dc2c88b517eb43323907b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa699a1e209131cea23030.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa5f3a1e209131ce9e8680.md>)

[Implementation](<https://github.com/rgeirhos/texture-vs-shape>)

Recorded evaluation configuration: {"Backbone":"ResNet-50"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.