# Image Classification on Red Mini-ImageNet (test) (Noise Robustness)

Source: [Image Classification on Red Mini-ImageNet (test) (Noise Robustness)](<https://www.sota2.com/research/sota/image-classification-on-red-mini-imagenet-test-noise-robustness>)

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

Data updated: 2026-02-26T05:34:21.716Z

Evaluation of Image Classification performance on the test split of Red Mini-ImageNet under varying levels of data corruption (noise rates 0.2, 0.4, 0.6, and 0.8).

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) | Red Mini-ImageNet (test) |
| Metric | Accuracy (Noise 0.2) |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 16 of 16 |
| 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 (Noise 0.2)](<https://www.sota2.com/research/sota/image-classification-on-red-mini-imagenet-test-noise-robustness.md?metric=Accuracy%20(Noise%200.2)>): Higher is better; unit: PERCENT.

- [Accuracy (Noise 0.4)](<https://www.sota2.com/research/sota/image-classification-on-red-mini-imagenet-test-noise-robustness.md?metric=Accuracy%20(Noise%200.4)>): Higher is better; unit: PERCENT.

- [Accuracy (Noise 0.6)](<https://www.sota2.com/research/sota/image-classification-on-red-mini-imagenet-test-noise-robustness.md?metric=Accuracy%20(Noise%200.6)>): Higher is better; unit: PERCENT.

- [Accuracy (Noise 0.8)](<https://www.sota2.com/research/sota/image-classification-on-red-mini-imagenet-test-noise-robustness.md?metric=Accuracy%20(Noise%200.8)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy (Noise 0.2) | Result date | Report paper |
| --- | --- | --- | --- |
| Ours (CLIPCleaner + MixFix) | 61.44 | Not recorded | 68fa74db1e209131cea6432a |
| GNL - p(Y\|X) ensemble | 58.46 | Not recorded | 68fa72531e209131cea55a73 |
| InstanceGM | 58.38 | Not recorded | 68fa74db1e209131cea6432a |
| GNL - p(X\|Y) ensemble | 57.56 | Not recorded | 68fa72531e209131cea55a73 |
| GNL - p(Y\|X) | 54.53 | Not recorded | 68fa72531e209131cea55a73 |
| GNL - p(X\|Y) | 53.34 | Not recorded | 68fa72531e209131cea55a73 |
| FaMUS | 51.42 | Not recorded | 68fa72531e209131cea55a73 |
| FaMUS | 51.42 | Not recorded | 68fa74db1e209131cea6432a |
| MentorMix | 51.02 | Not recorded | 68fa72531e209131cea55a73 |
| MentorMix | 51.02 | Not recorded | 68fa74db1e209131cea6432a |
| DivideMix | 50.96 | Not recorded | 68fa72531e209131cea55a73 |
| DivideMix | 50.96 | Not recorded | 68fa74db1e209131cea6432a |
| Mixup | 49.1 | Not recorded | 68fa72531e209131cea55a73 |
| Mixup | 49.1 | Not recorded | 68fa74db1e209131cea6432a |
| CE | 47.36 | Not recorded | 68fa72531e209131cea55a73 |
| CE | 47.36 | Not recorded | 68fa74db1e209131cea6432a |

### Ours (CLIPCleaner + MixFix)

Accuracy (Noise 0.2): 61.44. Result ID: 69937d7d3a6c314849165054.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74db1e209131cea6432a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74db1e209131cea6432a.md>)

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

### GNL - p(Y\|X) ensemble

Accuracy (Noise 0.2): 58.46. Result ID: 699230913a6c3148490d2f03.

[Reporting paper](<https://www.sota2.com/research/paper/68fa72531e209131cea55a73.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72531e209131cea55a73.md>)

[Implementation](<https://github.com/lfb-1/GNL>)

Recorded evaluation configuration: {"Ensemble":true}. This record does not assert independent reproduction.

### InstanceGM

Accuracy (Noise 0.2): 58.38. Result ID: 69937d7d3a6c314849165053.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74db1e209131cea6432a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d591e209131cea395bc.md>)

[Implementation](<https://github.com/a5507203/IDLN/blob/main/causalNL.py>)

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

### GNL - p(X\|Y) ensemble

Accuracy (Noise 0.2): 57.56. Result ID: 699230913a6c3148490d2f04.

[Reporting paper](<https://www.sota2.com/research/paper/68fa72531e209131cea55a73.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72531e209131cea55a73.md>)

[Implementation](<https://github.com/lfb-1/GNL>)

Recorded evaluation configuration: {"Ensemble":true}. This record does not assert independent reproduction.

### GNL - p(Y\|X)

Accuracy (Noise 0.2): 54.53. Result ID: 699230913a6c3148490d2efe.

[Reporting paper](<https://www.sota2.com/research/paper/68fa72531e209131cea55a73.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72531e209131cea55a73.md>)

[Implementation](<https://github.com/lfb-1/GNL>)

Recorded evaluation configuration: {"Ensemble":false}. This record does not assert independent reproduction.

### GNL - p(X\|Y)

Accuracy (Noise 0.2): 53.34. Result ID: 699230913a6c3148490d2eff.

[Reporting paper](<https://www.sota2.com/research/paper/68fa72531e209131cea55a73.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72531e209131cea55a73.md>)

[Implementation](<https://github.com/lfb-1/GNL>)

Recorded evaluation configuration: {"Ensemble":false}. This record does not assert independent reproduction.

### FaMUS

Accuracy (Noise 0.2): 51.42. Result ID: 699230913a6c3148490d2f02.

[Reporting paper](<https://www.sota2.com/research/paper/68fa72531e209131cea55a73.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69d11e209131cea245ef.md>)

[Implementation](<https://github.com/youjiangxu/FaMUS>)

Recorded evaluation configuration: {"Ensemble":false}. This record does not assert independent reproduction.

### FaMUS

Accuracy (Noise 0.2): 51.42. Result ID: 69937d7d3a6c314849165052.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74db1e209131cea6432a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69d11e209131cea245ef.md>)

[Implementation](<https://github.com/youjiangxu/FaMUS>)

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

### MentorMix

Accuracy (Noise 0.2): 51.02. Result ID: 699230913a6c3148490d2f01.

[Reporting paper](<https://www.sota2.com/research/paper/68fa72531e209131cea55a73.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa68321e209131cea1aab4.md>)

[Implementation](<http://www.lujiang.info/cnlw.html>)

Recorded evaluation configuration: {"Ensemble":false}. This record does not assert independent reproduction.

### MentorMix

Accuracy (Noise 0.2): 51.02. Result ID: 69937d7d3a6c314849165051.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74db1e209131cea6432a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa68321e209131cea1aab4.md>)

[Implementation](<http://www.lujiang.info/cnlw.html>)

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

### DivideMix

Accuracy (Noise 0.2): 50.96. Result ID: 699230913a6c3148490d2f00.

[Reporting paper](<https://www.sota2.com/research/paper/68fa72531e209131cea55a73.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa671c1e209131cea144cf.md>)

[Implementation](<https://github.com/LiJunnan1992/DivideMix>)

Recorded evaluation configuration: {"Ensemble":false}. This record does not assert independent reproduction.

### DivideMix

Accuracy (Noise 0.2): 50.96. Result ID: 69937d7d3a6c314849165050.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74db1e209131cea6432a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa671c1e209131cea144cf.md>)

[Implementation](<https://github.com/LiJunnan1992/DivideMix>)

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

### Mixup

Accuracy (Noise 0.2): 49.1. Result ID: 699230913a6c3148490d2efd.

[Reporting paper](<https://www.sota2.com/research/paper/68fa72531e209131cea55a73.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa644e1e209131cea03e44.md>)

[Implementation](<https://github.com/facebookresearch/mixup-cifar10>)

Recorded evaluation configuration: {"Ensemble":false}. This record does not assert independent reproduction.

### Mixup

Accuracy (Noise 0.2): 49.1. Result ID: 69937d7d3a6c31484916504f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74db1e209131cea6432a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa644e1e209131cea03e44.md>)

[Implementation](<https://github.com/facebookresearch/mixup-cifar10>)

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

### CE

Accuracy (Noise 0.2): 47.36. Result ID: 699230913a6c3148490d2efc.

[Reporting paper](<https://www.sota2.com/research/paper/68fa72531e209131cea55a73.md>)

Recorded evaluation configuration: {"Ensemble":false}. This record does not assert independent reproduction.

### CE

Accuracy (Noise 0.2): 47.36. Result ID: 69937d7d3a6c31484916504e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74db1e209131cea6432a.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.