# Image Classification on CIFAR-10 Severity 2 (Clean and PGD-20 Robustness)

Source: [Image Classification on CIFAR-10 Severity 2 (Clean and PGD-20 Robustness)](<https://www.sota2.com/research/sota/image-classification-on-cifar-10-severity-2-clean-and-pgd-20-robustness>)

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

Data updated: 2026-04-14T12:21:58.442Z

Evaluation of Image Classification performance on the CIFAR-10 Severity 2 dataset, reporting both clean accuracy and robustness against Projected Gradient Descent (PGD-20) attacks.

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 | Image Classification |
| Dataset (including any stated split) | CIFAR-10 Severity 2 |
| Metric | Clean Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 6 of 6 |
| 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

- [Clean Accuracy](<https://www.sota2.com/research/sota/image-classification-on-cifar-10-severity-2-clean-and-pgd-20-robustness.md?metric=Clean%20Accuracy>): Higher is better; unit: PERCENT.

- [Robust Accuracy (PGD-20)](<https://www.sota2.com/research/sota/image-classification-on-cifar-10-severity-2-clean-and-pgd-20-robustness.md?metric=Robust%20Accuracy%20(PGD-20)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Clean Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| PGD | 66.51 | Not recorded | 69ddca7b7d0172140ec79f75 |
| TgRA | 62.1 | Not recorded | 69ddca7b7d0172140ec79f75 |
| DKL | 60.04 | Not recorded | 69ddca7b7d0172140ec79f75 |
| TRADES-U | 54.05 | Not recorded | 69ddca7b7d0172140ec79f75 |
| TRADES | 48.43 | Not recorded | 69ddca7b7d0172140ec79f75 |
| MART | 30.69 | Not recorded | 69ddca7b7d0172140ec79f75 |

### PGD

Clean Accuracy: 66.51. Result ID: 69de314847f34ce924c6faea.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca7b7d0172140ec79f75.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa665a1e209131cea0fbae.md>)

Recorded evaluation configuration: {"Backbone":"WRNet-34"}. This record does not assert independent reproduction.

### TgRA

Clean Accuracy: 62.1. Result ID: 69de314847f34ce924c6faef.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca7b7d0172140ec79f75.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69ddca7b7d0172140ec79f75.md>)

Recorded evaluation configuration: {"Backbone":"WRNet-34"}. This record does not assert independent reproduction.

### DKL

Clean Accuracy: 60.04. Result ID: 69de314847f34ce924c6faed.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca7b7d0172140ec79f75.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa756d1e209131cea67d56.md>)

[Implementation](<https://github.com/jiequancui/DKL>)

Recorded evaluation configuration: {"Backbone":"WRNet-34"}. This record does not assert independent reproduction.

### TRADES-U

Clean Accuracy: 54.05. Result ID: 69de314847f34ce924c6faee.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca7b7d0172140ec79f75.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65f71e209131cea0d897.md>)

Recorded evaluation configuration: {"Backbone":"WRNet-34"}. This record does not assert independent reproduction.

### TRADES

Clean Accuracy: 48.43. Result ID: 69de314847f34ce924c6faeb.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca7b7d0172140ec79f75.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65f71e209131cea0d897.md>)

Recorded evaluation configuration: {"Backbone":"WRNet-34"}. This record does not assert independent reproduction.

### MART

Clean Accuracy: 30.69. Result ID: 69de314847f34ce924c6faec.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca7b7d0172140ec79f75.md>)

Recorded evaluation configuration: {"Backbone":"WRNet-34"}. This record does not assert independent reproduction.

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