# Adversarial Robustness Accuracy on SVHN (FGSM, PGD, BIM, APGD, Jitter)

Source: [Adversarial Robustness Accuracy on SVHN (FGSM, PGD, BIM, APGD, Jitter)](<https://www.sota2.com/research/sota/adversarial-robustness-accuracy-on-svhn-fgsm-pgd-bim-apgd-jitter>)

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

Data updated: 2026-04-14T10:34:32.581Z

Evaluation of model robustness on the SVHN dataset against various adversarial attacks, including FGSM, PGD, BIM, APGD, and Jitter, reporting the resulting accuracy for each perturbation level.

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 | Robustness Acc (FGSM, ε=8/255) |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 7 of 7 |
| 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

- [Robustness Acc (FGSM, ε=8/255)](<https://www.sota2.com/research/sota/adversarial-robustness-accuracy-on-svhn-fgsm-pgd-bim-apgd-jitter.md?metric=Robustness%20Acc%20(FGSM%2C%20%CE%B5%3D8%2F255)>): Higher is better; unit: PERCENT.

- [Robustness Acc (FGSM, ε=16/255)](<https://www.sota2.com/research/sota/adversarial-robustness-accuracy-on-svhn-fgsm-pgd-bim-apgd-jitter.md?metric=Robustness%20Acc%20(FGSM%2C%20%CE%B5%3D16%2F255)>): Higher is better; unit: PERCENT.

- [Robustness Acc (PGD, ε=8/255)](<https://www.sota2.com/research/sota/adversarial-robustness-accuracy-on-svhn-fgsm-pgd-bim-apgd-jitter.md?metric=Robustness%20Acc%20(PGD%2C%20%CE%B5%3D8%2F255)>): Higher is better; unit: PERCENT.

- [Robustness Acc (PGD, ε=16/255)](<https://www.sota2.com/research/sota/adversarial-robustness-accuracy-on-svhn-fgsm-pgd-bim-apgd-jitter.md?metric=Robustness%20Acc%20(PGD%2C%20%CE%B5%3D16%2F255)>): Higher is better; unit: PERCENT.

- [Robustness Acc (BIM, ε=8/255)](<https://www.sota2.com/research/sota/adversarial-robustness-accuracy-on-svhn-fgsm-pgd-bim-apgd-jitter.md?metric=Robustness%20Acc%20(BIM%2C%20%CE%B5%3D8%2F255)>): Higher is better; unit: PERCENT.

- [Robustness Acc (BIM, ε=16/255)](<https://www.sota2.com/research/sota/adversarial-robustness-accuracy-on-svhn-fgsm-pgd-bim-apgd-jitter.md?metric=Robustness%20Acc%20(BIM%2C%20%CE%B5%3D16%2F255)>): Higher is better; unit: PERCENT.

- [Robustness Acc (APGD, ε=8/255)](<https://www.sota2.com/research/sota/adversarial-robustness-accuracy-on-svhn-fgsm-pgd-bim-apgd-jitter.md?metric=Robustness%20Acc%20(APGD%2C%20%CE%B5%3D8%2F255)>): Higher is better; unit: PERCENT.

- [Robustness Acc (APGD, ε=16/255)](<https://www.sota2.com/research/sota/adversarial-robustness-accuracy-on-svhn-fgsm-pgd-bim-apgd-jitter.md?metric=Robustness%20Acc%20(APGD%2C%20%CE%B5%3D16%2F255)>): Higher is better; unit: PERCENT.

- [Robustness Acc (Jitter, ε=8/255)](<https://www.sota2.com/research/sota/adversarial-robustness-accuracy-on-svhn-fgsm-pgd-bim-apgd-jitter.md?metric=Robustness%20Acc%20(Jitter%2C%20%CE%B5%3D8%2F255)>): Higher is better; unit: PERCENT.

- [Robustness Acc (Jitter, ε=16/255)](<https://www.sota2.com/research/sota/adversarial-robustness-accuracy-on-svhn-fgsm-pgd-bim-apgd-jitter.md?metric=Robustness%20Acc%20(Jitter%2C%20%CE%B5%3D16%2F255)>): Higher is better; unit: PERCENT.

- [Avg Robustness Acc](<https://www.sota2.com/research/sota/adversarial-robustness-accuracy-on-svhn-fgsm-pgd-bim-apgd-jitter.md?metric=Avg%20Robustness%20Acc>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Robustness Acc (FGSM, ε=8/255) | Result date | Report paper |
| --- | --- | --- | --- |
| Zubov-Net | 71.06 | Not recorded | 68fa5e551e209131ce9e165d |
| FxTS-Net | 66.1 | Not recorded | 68fa5e551e209131ce9e165d |
| LyaNet | 65.26 | Not recorded | 68fa5e551e209131ce9e165d |
| Proj-NODE | 65.04 | Not recorded | 68fa5e551e209131ce9e165d |
| SODEF | 64.09 | Not recorded | 68fa5e551e209131ce9e165d |
| Neural ODE | 63.86 | Not recorded | 68fa5e551e209131ce9e165d |
| ResNet18 | 62.96 | Not recorded | 68fa5e551e209131ce9e165d |

### Zubov-Net

Robustness Acc (FGSM, ε=8/255): 71.06. Result ID: 69de183147f34ce924c63c74.

[Reporting paper](<https://www.sota2.com/research/paper/68fa5e551e209131ce9e165d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa5e551e209131ce9e165d.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### FxTS-Net

Robustness Acc (FGSM, ε=8/255): 66.1. Result ID: 69de183147f34ce924c63c72.

[Reporting paper](<https://www.sota2.com/research/paper/68fa5e551e209131ce9e165d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa5c991e209131ce9d27a1.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### LyaNet

Robustness Acc (FGSM, ε=8/255): 65.26. Result ID: 69de183147f34ce924c63c73.

[Reporting paper](<https://www.sota2.com/research/paper/68fa5e551e209131ce9e165d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6bc71e209131cea30449.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Proj-NODE

Robustness Acc (FGSM, ε=8/255): 65.04. Result ID: 69de183147f34ce924c63c70.

[Reporting paper](<https://www.sota2.com/research/paper/68fa5e551e209131ce9e165d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa56921e209131ce99dbee.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### SODEF

Robustness Acc (FGSM, ε=8/255): 64.09. Result ID: 69de183147f34ce924c63c71.

[Reporting paper](<https://www.sota2.com/research/paper/68fa5e551e209131ce9e165d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6b1b1e209131cea2c3f2.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Neural ODE

Robustness Acc (FGSM, ε=8/255): 63.86. Result ID: 69de183147f34ce924c63c6f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa5e551e209131ce9e165d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa66cb1e209131cea12502.md>)

[Implementation](<https://www.sota2.com/github.com/rtqichen/torchdiffeq>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### ResNet18

Robustness Acc (FGSM, ε=8/255): 62.96. Result ID: 69de183147f34ce924c63c6e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa5e551e209131ce9e165d.md>)

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

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

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