# Image Classification on CIFAR-100 (FGSM Error)

Source: [Image Classification on CIFAR-100 (FGSM Error)](<https://www.sota2.com/research/sota/image-classification-on-cifar-100-fgsm-error>)

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

Data updated: 2026-02-27T07:28:40.257Z

Evaluation of Image Classification performance on the CIFAR-100 dataset, specifically measuring adversarial robustness using the Fast Gradient Sign Method (FGSM) Error rate.

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-100 |
| Metric | FGSM Error Rate (%) |
| Unit | PERCENT |
| Direction | Lower 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

- [FGSM Error Rate (%)](<https://www.sota2.com/research/sota/image-classification-on-cifar-100-fgsm-error.md?metric=FGSM%20Error%20Rate%20(%25)>): Lower is better; unit: PERCENT.

## Recorded results

| Method | FGSM Error Rate (%) | Result date | Report paper |
| --- | --- | --- | --- |
| AutoMix | 55.34 | Not recorded | 68fa6d701e209131cea39e02 |
| AugMix | 55.59 | Not recorded | 68fa6d701e209131cea39e02 |
| MixUp | 56.6 | Not recorded | 68fa6d701e209131cea39e02 |
| PuzzleMix | 63.71 | Not recorded | 68fa6d701e209131cea39e02 |
| Vanilla | 63.92 | Not recorded | 68fa6d701e209131cea39e02 |
| CutMix | 76.84 | Not recorded | 68fa6d701e209131cea39e02 |

### AutoMix

FGSM Error Rate (%): 55.34. Result ID: 69a147a335dfcabb00e5fb5e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6d701e209131cea39e02.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d701e209131cea39e02.md>)

[Implementation](<https://github.com/Westlake-AI/openmixup>)

Recorded evaluation configuration: {"Backbone":"ResNeXt-50 (32x4d)","Training Epochs":"400","Attack Method":"FGSM","Attack Type":"white-box","Epsilon":"8/255 l\_inf"}. This record does not assert independent reproduction.

### AugMix

FGSM Error Rate (%): 55.59. Result ID: 69a147a335dfcabb00e5fb5c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6d701e209131cea39e02.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa67041e209131cea13ccb.md>)

[Implementation](<https://github.com/google-research/augmix>)

Recorded evaluation configuration: {"Backbone":"ResNeXt-50 (32x4d)","Training Epochs":"400","Attack Method":"FGSM","Attack Type":"white-box","Epsilon":"8/255 l\_inf"}. This record does not assert independent reproduction.

### MixUp

FGSM Error Rate (%): 56.6. Result ID: 69a147a335dfcabb00e5fb5a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6d701e209131cea39e02.md>)

Recorded evaluation configuration: {"Backbone":"ResNeXt-50 (32x4d)","Training Epochs":"400","Attack Method":"FGSM","Attack Type":"white-box","Epsilon":"8/255 l\_inf"}. This record does not assert independent reproduction.

### PuzzleMix

FGSM Error Rate (%): 63.71. Result ID: 69a147a335dfcabb00e5fb5d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6d701e209131cea39e02.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa68f91e209131cea1f6ad.md>)

Recorded evaluation configuration: {"Backbone":"ResNeXt-50 (32x4d)","Training Epochs":"400","Attack Method":"FGSM","Attack Type":"white-box","Epsilon":"8/255 l\_inf"}. This record does not assert independent reproduction.

### Vanilla

FGSM Error Rate (%): 63.92. Result ID: 69a147a335dfcabb00e5fb59.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6d701e209131cea39e02.md>)

Recorded evaluation configuration: {"Backbone":"ResNeXt-50 (32x4d)","Training Epochs":"400","Attack Method":"FGSM","Attack Type":"white-box","Epsilon":"8/255 l\_inf"}. This record does not assert independent reproduction.

### CutMix

FGSM Error Rate (%): 76.84. Result ID: 69a147a335dfcabb00e5fb5b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6d701e209131cea39e02.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa663d1e209131cea0f1ce.md>)

[Implementation](<https://github.com/clovaai/CutMix-PyTorch>)

Recorded evaluation configuration: {"Backbone":"ResNeXt-50 (32x4d)","Training Epochs":"400","Attack Method":"FGSM","Attack Type":"white-box","Epsilon":"8/255 l\_inf"}. This record does not assert independent reproduction.

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