# Image Classification on CIFAR-10 (test) (Noise Accuracy Profile)

Source: [Image Classification on CIFAR-10 (test) (Noise Accuracy Profile)](<https://www.sota2.com/research/sota/image-classification-on-cifar-10-test-noise-accuracy-profile>)

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

Data updated: 2026-02-26T05:34:27.727Z

Evaluation of Image Classification performance on the CIFAR-10 test set, measuring accuracy under varying levels of symmetric and asymmetric data corruption/noise injection (e.g., 20%, 40%, 90% noise levels).

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 (Symmetric 20% Noise) |
| 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

- [Accuracy (Symmetric 20% Noise)](<https://www.sota2.com/research/sota/image-classification-on-cifar-10-test-noise-accuracy-profile.md?metric=Accuracy%20(Symmetric%2020%25%20Noise)>): Higher is better; unit: PERCENT.

- [Accuracy (Symmetric 90% Noise)](<https://www.sota2.com/research/sota/image-classification-on-cifar-10-test-noise-accuracy-profile.md?metric=Accuracy%20(Symmetric%2090%25%20Noise)>): Higher is better; unit: PERCENT.

- [Accuracy (Asymmetric 40% Noise)](<https://www.sota2.com/research/sota/image-classification-on-cifar-10-test-noise-accuracy-profile.md?metric=Accuracy%20(Asymmetric%2040%25%20Noise)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy (Symmetric 20% Noise) | Result date | Report paper |
| --- | --- | --- | --- |
| NGC | 95.9 | Not recorded | 68fa6bf21e209131cea31141 |
| ProtoMix | 95.8 | Not recorded | 68fa6bf21e209131cea31141 |
| Sel-CL+ | 95.4 | Not recorded | 68fa6bf21e209131cea31141 |
| Sel-CL+ | 95.2 | Not recorded | 68fa6bf21e209131cea31141 |

### NGC

Accuracy (Symmetric 20% Noise): 95.9. Result ID: 698eab8cffc3901de09fa6c7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6bf21e209131cea31141.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6a951e209131cea2911d.md>)

Recorded evaluation configuration: {"Training Pipeline":"One-stage"}. This record does not assert independent reproduction.

### ProtoMix

Accuracy (Symmetric 20% Noise): 95.8. Result ID: 698eab8cffc3901de09fa6c6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6bf21e209131cea31141.md>)

Recorded evaluation configuration: {"Training Pipeline":"One-stage"}. This record does not assert independent reproduction.

### Sel-CL+

Accuracy (Symmetric 20% Noise): 95.4. Result ID: 698eab8cffc3901de09fa6c8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6bf21e209131cea31141.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6bf21e209131cea31141.md>)

[Implementation](<https://github.com/ShikunLi/Sel-CL>)

Recorded evaluation configuration: {"Training Pipeline":"One-stage","Fine-tuning Augmentation":"Standard"}. This record does not assert independent reproduction.

### Sel-CL+

Accuracy (Symmetric 20% Noise): 95.2. Result ID: 698eab8cffc3901de09fa6c9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6bf21e209131cea31141.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6bf21e209131cea31141.md>)

[Implementation](<https://github.com/ShikunLi/Sel-CL>)

Recorded evaluation configuration: {"Training Pipeline":"One-stage","Fine-tuning Augmentation":"AugMix"}. This record does not assert independent reproduction.

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