# Image Classification Robustness on STL-10 (Adversarial Attacks)

Source: [Image Classification Robustness on STL-10 (Adversarial Attacks)](<https://www.sota2.com/research/sota/image-classification-robustness-on-stl-10-adversarial-attacks>)

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

Data updated: 2026-02-27T11:19:51.974Z

Evaluation of Image Classification performance on the STL-10 dataset focusing on robustness against various adversarial attacks, including FGSM, PGD, SparseFool, and PAP.

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) | STL-10 |
| Metric | Accuracy (Benign) |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 14 of 14 |
| 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 (Benign)](<https://www.sota2.com/research/sota/image-classification-robustness-on-stl-10-adversarial-attacks.md?metric=Accuracy%20(Benign)>): Higher is better; unit: PERCENT.

- [Accuracy (FGSM)](<https://www.sota2.com/research/sota/image-classification-robustness-on-stl-10-adversarial-attacks.md?metric=Accuracy%20(FGSM)>): Higher is better; unit: PERCENT.

- [Accuracy (PGD)](<https://www.sota2.com/research/sota/image-classification-robustness-on-stl-10-adversarial-attacks.md?metric=Accuracy%20(PGD)>): Higher is better; unit: PERCENT.

- [Accuracy (SparseFool)](<https://www.sota2.com/research/sota/image-classification-robustness-on-stl-10-adversarial-attacks.md?metric=Accuracy%20(SparseFool)>): Higher is better; unit: PERCENT.

- [Accuracy (PAP)](<https://www.sota2.com/research/sota/image-classification-robustness-on-stl-10-adversarial-attacks.md?metric=Accuracy%20(PAP)>): Higher is better; unit: PERCENT.

- [Robust Accuracy](<https://www.sota2.com/research/sota/image-classification-robustness-on-stl-10-adversarial-attacks.md?metric=Robust%20Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy (Benign) | Result date | Report paper |
| --- | --- | --- | --- |
| Lorica | 49.9 | Not recorded | 68fa792a1e209131cea7c5d5 |
| Sylva | 48.21 | Not recorded | 68fa792a1e209131cea7c5d5 |
| FP-Gen-AF | 43.1 | Not recorded | 68fa792a1e209131cea7c5d5 |
| Per-LoRA | 43.06 | Not recorded | 68fa792a1e209131cea7c5d5 |
| DBFAT | 42.75 | Not recorded | 68fa792a1e209131cea7c5d5 |
| FA-Gen-AF | 42.29 | Not recorded | 68fa792a1e209131cea7c5d5 |
| FP-TRADES | 41.1 | Not recorded | 68fa792a1e209131cea7c5d5 |
| FA-TRADES | 40.31 | Not recorded | 68fa792a1e209131cea7c5d5 |
| FP-PGD-AT | 34.41 | Not recorded | 68fa792a1e209131cea7c5d5 |
| Per-Adv | 32.92 | Not recorded | 68fa792a1e209131cea7c5d5 |
| FA-PGD-AT | 32.81 | Not recorded | 68fa792a1e209131cea7c5d5 |

### Lorica

Accuracy (Benign): 49.9. Result ID: 699a8b725698665ec415a0e0.

[Reporting paper](<https://www.sota2.com/research/paper/68fa792a1e209131cea7c5d5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa792a1e209131cea7c5d5.md>)

Recorded evaluation configuration: {"Backbone":"ViT-T/16"}. This record does not assert independent reproduction.

### Sylva

Accuracy (Benign): 48.21. Result ID: 699a8b725698665ec415a0df.

[Reporting paper](<https://www.sota2.com/research/paper/68fa792a1e209131cea7c5d5.md>)

Recorded evaluation configuration: {"Backbone":"ViT-T/16"}. This record does not assert independent reproduction.

### FP-Gen-AF

Accuracy (Benign): 43.1. Result ID: 699a8b725698665ec415a0db.

[Reporting paper](<https://www.sota2.com/research/paper/68fa792a1e209131cea7c5d5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa728e1e209131cea570cd.md>)

Recorded evaluation configuration: {"Federated Strategy":"FedProx","Local Defense":"Gen-AF","Backbone":"ViT-T/16"}. This record does not assert independent reproduction.

### Per-LoRA

Accuracy (Benign): 43.06. Result ID: 699a8b725698665ec415a0de.

[Reporting paper](<https://www.sota2.com/research/paper/68fa792a1e209131cea7c5d5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74601e209131cea6146b.md>)

Recorded evaluation configuration: {"Backbone":"ViT-T/16"}. This record does not assert independent reproduction.

### DBFAT

Accuracy (Benign): 42.75. Result ID: 699a8b725698665ec415a0dc.

[Reporting paper](<https://www.sota2.com/research/paper/68fa792a1e209131cea7c5d5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6e961e209131cea40e66.md>)

Recorded evaluation configuration: {"Backbone":"ViT-T/16"}. This record does not assert independent reproduction.

### FA-Gen-AF

Accuracy (Benign): 42.29. Result ID: 699a8b725698665ec415a0d8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa792a1e209131cea7c5d5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa728e1e209131cea570cd.md>)

Recorded evaluation configuration: {"Federated Strategy":"FedAvg","Local Defense":"Gen-AF","Backbone":"ViT-T/16"}. This record does not assert independent reproduction.

### FP-TRADES

Accuracy (Benign): 41.1. Result ID: 699a8b725698665ec415a0da.

[Reporting paper](<https://www.sota2.com/research/paper/68fa792a1e209131cea7c5d5.md>)

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

Recorded evaluation configuration: {"Federated Strategy":"FedProx","Local Defense":"TRADES","Backbone":"ViT-T/16"}. This record does not assert independent reproduction.

### FA-TRADES

Accuracy (Benign): 40.31. Result ID: 699a8b725698665ec415a0d7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa792a1e209131cea7c5d5.md>)

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

Recorded evaluation configuration: {"Federated Strategy":"FedAvg","Local Defense":"TRADES","Backbone":"ViT-T/16"}. This record does not assert independent reproduction.

### FP-PGD-AT

Accuracy (Benign): 34.41. Result ID: 699a8b725698665ec415a0d9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa792a1e209131cea7c5d5.md>)

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

Recorded evaluation configuration: {"Federated Strategy":"FedProx","Local Defense":"PGD-AT","Backbone":"ViT-T/16"}. This record does not assert independent reproduction.

### Per-Adv

Accuracy (Benign): 32.92. Result ID: 699a8b725698665ec415a0dd.

[Reporting paper](<https://www.sota2.com/research/paper/68fa792a1e209131cea7c5d5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75051e209131cea654c5.md>)

Recorded evaluation configuration: {"Backbone":"ViT-T/16"}. This record does not assert independent reproduction.

### FA-PGD-AT

Accuracy (Benign): 32.81. Result ID: 699a8b725698665ec415a0d6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa792a1e209131cea7c5d5.md>)

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

Recorded evaluation configuration: {"Federated Strategy":"FedAvg","Local Defense":"PGD-AT","Backbone":"ViT-T/16"}. This record does not assert independent reproduction.

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