# Image Classification on CIFAR-10 (test) (Transformation Robustness)

Source: [Image Classification on CIFAR-10 (test) (Transformation Robustness)](<https://www.sota2.com/research/sota/image-classification-on-cifar-10-test-transformation-robustness>)

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

Data updated: 2026-02-26T05:34:10.210Z

Evaluation of Image Classification performance on the CIFAR-10 test set, focusing on accuracy under various geometric transformations (Rotation, Perspective) and data augmentation techniques (Cutout, MixUp).

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 (test) |
| Metric | Accuracy (Base) |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 11 of 11 |
| 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 (Base)](<https://www.sota2.com/research/sota/image-classification-on-cifar-10-test-transformation-robustness.md?metric=Accuracy%20(Base)>): Higher is better; unit: PERCENT.

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

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

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

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

- [Accuracy (Channel Shuffle)](<https://www.sota2.com/research/sota/image-classification-on-cifar-10-test-transformation-robustness.md?metric=Accuracy%20(Channel%20Shuffle)>): Higher is better; unit: PERCENT.

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

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

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

## Recorded results

| Method | Accuracy (Base) | Result date | Report paper |
| --- | --- | --- | --- |
| Clean | 83.93 | Not recorded | 697ae969a2b28e761d87cede |
| CUDA\* | 65.47 | Not recorded | 697ae969a2b28e761d87cede |
| TAP | 35.9 | Not recorded | 697ae969a2b28e761d87cede |
| SEP | 28.43 | Not recorded | 697ae969a2b28e761d87cede |
| NTGA | 27.6 | Not recorded | 697ae969a2b28e761d87cede |
| EM | 21.43 | Not recorded | 697ae969a2b28e761d87cede |
| REM | 21.36 | Not recorded | 697ae969a2b28e761d87cede |
| AR | 16.89 | Not recorded | 697ae969a2b28e761d87cede |
| SP | 16.06 | Not recorded | 697ae969a2b28e761d87cede |
| PIL | 14.7 | Not recorded | 697ae969a2b28e761d87cede |
| CUDA | 10.35 | Not recorded | 697ae969a2b28e761d87cede |

### Clean

Accuracy (Base): 83.93. Result ID: 69990c325698665ec471d185.

[Reporting paper](<https://www.sota2.com/research/paper/697ae969a2b28e761d87cede.md>)

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

### CUDA\*

Accuracy (Base): 65.47. Result ID: 69990c325698665ec471d18c.

[Reporting paper](<https://www.sota2.com/research/paper/697ae969a2b28e761d87cede.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6eb91e209131cea41ccf.md>)

[Implementation](<https://github.com/vinusankars/Convolution-based-Unlearnability>)

Recorded evaluation configuration: {"Perturbation budget (epsilon)":"8/255"}. This record does not assert independent reproduction.

### TAP

Accuracy (Base): 35.9. Result ID: 69990c325698665ec471d188.

[Reporting paper](<https://www.sota2.com/research/paper/697ae969a2b28e761d87cede.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6a3d1e209131cea26ecb.md>)

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

### SEP

Accuracy (Base): 28.43. Result ID: 69990c325698665ec471d18d.

[Reporting paper](<https://www.sota2.com/research/paper/697ae969a2b28e761d87cede.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f041e209131cea43464.md>)

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

### NTGA

Accuracy (Base): 27.6. Result ID: 69990c325698665ec471d18a.

[Reporting paper](<https://www.sota2.com/research/paper/697ae969a2b28e761d87cede.md>)

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

### EM

Accuracy (Base): 21.43. Result ID: 69990c325698665ec471d186.

[Reporting paper](<https://www.sota2.com/research/paper/697ae969a2b28e761d87cede.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69541e209131cea2194b.md>)

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

### REM

Accuracy (Base): 21.36. Result ID: 69990c325698665ec471d187.

[Reporting paper](<https://www.sota2.com/research/paper/697ae969a2b28e761d87cede.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c1c1e209131cea3246f.md>)

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

### AR

Accuracy (Base): 16.89. Result ID: 69990c325698665ec471d18e.

[Reporting paper](<https://www.sota2.com/research/paper/697ae969a2b28e761d87cede.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6dab1e209131cea3b59a.md>)

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

### SP

Accuracy (Base): 16.06. Result ID: 69990c325698665ec471d189.

[Reporting paper](<https://www.sota2.com/research/paper/697ae969a2b28e761d87cede.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6cb71e209131cea35891.md>)

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

### PIL

Accuracy (Base): 14.7. Result ID: 69990c325698665ec471d18f.

[Reporting paper](<https://www.sota2.com/research/paper/697ae969a2b28e761d87cede.md>)

[Method origin paper](<https://www.sota2.com/research/paper/697ae969a2b28e761d87cede.md>)

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

### CUDA

Accuracy (Base): 10.35. Result ID: 69990c325698665ec471d18b.

[Reporting paper](<https://www.sota2.com/research/paper/697ae969a2b28e761d87cede.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6eb91e209131cea41ccf.md>)

[Implementation](<https://github.com/vinusankars/Convolution-based-Unlearnability>)

Recorded evaluation configuration: {"Perturbation budget":"original larger budget"}. This record does not assert independent reproduction.

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