# Image Classification on SVHN (Common Corruptions and l\_inf Robustness)

Source: [Image Classification on SVHN (Common Corruptions and l\_inf Robustness)](<https://www.sota2.com/research/sota/image-classification-on-svhn-common-corruptions-and-l_inf-robustness>)

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

Data updated: 2026-02-27T08:00:05.746Z

Evaluation of Image Classification performance on the SVHN dataset under various common corruptions (e.g., Hue, Brightness, Rotation) and specific l\_infinity adversarial attack metrics.

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) | SVHN |
| Metric | Clean Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 4 of 4 |
| 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-svhn-common-corruptions-and-l_inf-robustness.md?metric=Clean%20Accuracy>): Higher is better; unit: PERCENT.

- [Accuracy (Hue)](<https://www.sota2.com/research/sota/image-classification-on-svhn-common-corruptions-and-l_inf-robustness.md?metric=Accuracy%20(Hue)>): Higher is better; unit: PERCENT.

- [Accuracy (Saturation)](<https://www.sota2.com/research/sota/image-classification-on-svhn-common-corruptions-and-l_inf-robustness.md?metric=Accuracy%20(Saturation)>): Higher is better; unit: PERCENT.

- [Accuracy (Rotation)](<https://www.sota2.com/research/sota/image-classification-on-svhn-common-corruptions-and-l_inf-robustness.md?metric=Accuracy%20(Rotation)>): Higher is better; unit: PERCENT.

- [Accuracy (Brightness)](<https://www.sota2.com/research/sota/image-classification-on-svhn-common-corruptions-and-l_inf-robustness.md?metric=Accuracy%20(Brightness)>): Higher is better; unit: PERCENT.

- [Accuracy (Contrast)](<https://www.sota2.com/research/sota/image-classification-on-svhn-common-corruptions-and-l_inf-robustness.md?metric=Accuracy%20(Contrast)>): Higher is better; unit: PERCENT.

- [Robustness Acc ($l\_\\infty$, 20-step)](<https://www.sota2.com/research/sota/image-classification-on-svhn-common-corruptions-and-l_inf-robustness.md?metric=Robustness%20Acc%20(%24l_%5Cinfty%24%2C%2020-step)>): Higher is better; unit: PERCENT.

- [Robustness Acc (Auto $l\_\\infty$)](<https://www.sota2.com/research/sota/image-classification-on-svhn-common-corruptions-and-l_inf-robustness.md?metric=Robustness%20Acc%20(Auto%20%24l_%5Cinfty%24)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Clean Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Normal\* | 95.4 | Not recorded | 68fa6ecc1e209131cea42343 |
| GAT-fs\* | 93.6 | Not recorded | 68fa6ecc1e209131cea42343 |
| GAT-f\* | 93.4 | Not recorded | 68fa6ecc1e209131cea42343 |
| Trades\*∞ | 90.3 | Not recorded | 68fa6ecc1e209131cea42343 |

### Normal\*

Clean Accuracy: 95.4. Result ID: 69a14ef635dfcabb00e6277e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6ecc1e209131cea42343.md>)

Recorded evaluation configuration: {"Training":"Normal"}. This record does not assert independent reproduction.

### GAT-fs\*

Clean Accuracy: 93.6. Result ID: 69a14ef635dfcabb00e62781.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6ecc1e209131cea42343.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ecc1e209131cea42343.md>)

[Implementation](<https://hsiung.cc/CARBEN/>)

Recorded evaluation configuration: {"Training":"GAT-fs"}. This record does not assert independent reproduction.

### GAT-f\*

Clean Accuracy: 93.4. Result ID: 69a14ef635dfcabb00e62780.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6ecc1e209131cea42343.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ecc1e209131cea42343.md>)

[Implementation](<https://hsiung.cc/CARBEN/>)

Recorded evaluation configuration: {"Training":"GAT-f"}. This record does not assert independent reproduction.

### Trades\*∞

Clean Accuracy: 90.3. Result ID: 69a14ef635dfcabb00e6277f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6ecc1e209131cea42343.md>)

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

Recorded evaluation configuration: {"Training":"Trades"}. This record does not assert independent reproduction.

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