# Image Classification on CIFAR-100 (val) (20-80% Noise Robustness)

Source: [Image Classification on CIFAR-100 (val) (20-80% Noise Robustness)](<https://www.sota2.com/research/sota/image-classification-on-cifar-100-val-20-80-percent-noise-robustness>)

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

Data updated: 2026-02-26T05:34:02.025Z

Evaluation of Image Classification performance on the validation split of the CIFAR-100 dataset, measuring accuracy under varying levels of input noise (20%, 40%, 60%, and 80%).

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 (val) |
| Metric | Accuracy (40% noise) |
| 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 (40% noise)](<https://www.sota2.com/research/sota/image-classification-on-cifar-100-val-20-80-percent-noise-robustness.md?metric=Accuracy%20(40%25%20noise)>): Higher is better; unit: PERCENT.

- [Accuracy (20% noise)](<https://www.sota2.com/research/sota/image-classification-on-cifar-100-val-20-80-percent-noise-robustness.md?metric=Accuracy%20(20%25%20noise)>): Higher is better; unit: PERCENT.

- [Accuracy (60% noise)](<https://www.sota2.com/research/sota/image-classification-on-cifar-100-val-20-80-percent-noise-robustness.md?metric=Accuracy%20(60%25%20noise)>): Higher is better; unit: PERCENT.

- [Accuracy (80% noise)](<https://www.sota2.com/research/sota/image-classification-on-cifar-100-val-20-80-percent-noise-robustness.md?metric=Accuracy%20(80%25%20noise)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy (40% noise) | Result date | Report paper |
| --- | --- | --- | --- |
| MentorMix | 71.3 | Not recorded | 68fa68321e209131cea1aab4 |
| MD-DYR-SH | 70.1 | Not recorded | 68fa65d71e209131cea0cad4 |
| Arazo et al. | 70.1 | Not recorded | 68fa68321e209131cea1aab4 |
| Huang et al. | 69.2 | Not recorded | 68fa68321e209131cea1aab4 |
| MentorNet | 68.5 | Not recorded | 68fa68321e209131cea1aab4 |
| Jiang et al. (2018b) | 68 | Not recorded | 68fa65d71e209131cea0cad4 |
| Mixup | 66.8 | Not recorded | 68fa68321e209131cea1aab4 |
| M-DYR-H | 64.4 | Not recorded | 68fa65d71e209131cea0cad4 |
| Zhang & Sabuncu | 62.6 | Not recorded | 68fa68321e209131cea1aab4 |
| Ren et al. (2018) | 61.3 | Not recorded | 68fa65d71e209131cea0cad4 |
| Ma et al. (2018) | 52 | Not recorded | 68fa65d71e209131cea0cad4 |

### MentorMix

Accuracy (40% noise): 71.3. Result ID: 698d61958b517eb43321121b.

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

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

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

Recorded evaluation configuration: {"Backbone":"ResNet-32"}. This record does not assert independent reproduction.

### MD-DYR-SH

Accuracy (40% noise): 70.1. Result ID: 698ccc368b517eb4331cd019.

[Reporting paper](<https://www.sota2.com/research/paper/68fa65d71e209131cea0cad4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65d71e209131cea0cad4.md>)

[Implementation](<https://git.io/fjsvE>)

Recorded evaluation configuration: {"Architecture":"PRN-18"}. This record does not assert independent reproduction.

### Arazo et al.

Accuracy (40% noise): 70.1. Result ID: 698d61958b517eb433211216.

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

[Method origin paper](<https://www.sota2.com/research/paper/68fa65d71e209131cea0cad4.md>)

[Implementation](<https://git.io/fjsvE>)

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

### Huang et al.

Accuracy (40% noise): 69.2. Result ID: 698d61958b517eb43321121a.

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

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

### MentorNet

Accuracy (40% noise): 68.5. Result ID: 698d61958b517eb433211218.

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

[Method origin paper](<https://www.sota2.com/research/paper/68fa64bf1e209131cea063ec.md>)

[Implementation](<https://github.com/google/mentornet>)

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

### Jiang et al. (2018b)

Accuracy (40% noise): 68. Result ID: 698ccc368b517eb4331cd015.

[Reporting paper](<https://www.sota2.com/research/paper/68fa65d71e209131cea0cad4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64bf1e209131cea063ec.md>)

[Implementation](<https://github.com/google/mentornet>)

Recorded evaluation configuration: {"Architecture":"WRN-101"}. This record does not assert independent reproduction.

### Mixup

Accuracy (40% noise): 66.8. Result ID: 698d61958b517eb433211219.

[Reporting paper](<https://www.sota2.com/research/paper/68fa68321e209131cea1aab4.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.

### M-DYR-H

Accuracy (40% noise): 64.4. Result ID: 698ccc368b517eb4331cd018.

[Reporting paper](<https://www.sota2.com/research/paper/68fa65d71e209131cea0cad4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65d71e209131cea0cad4.md>)

[Implementation](<https://git.io/fjsvE>)

Recorded evaluation configuration: {"Architecture":"PRN-18"}. This record does not assert independent reproduction.

### Zhang & Sabuncu

Accuracy (40% noise): 62.6. Result ID: 698d61958b517eb433211217.

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

[Method origin paper](<https://www.sota2.com/research/paper/68fa651f1e209131cea088eb.md>)

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

### Ren et al. (2018)

Accuracy (40% noise): 61.3. Result ID: 698ccc368b517eb4331cd017.

[Reporting paper](<https://www.sota2.com/research/paper/68fa65d71e209131cea0cad4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65be1e209131cea0c1af.md>)

Recorded evaluation configuration: {"Architecture":"WRN-28"}. This record does not assert independent reproduction.

### Ma et al. (2018)

Accuracy (40% noise): 52. Result ID: 698ccc368b517eb4331cd016.

[Reporting paper](<https://www.sota2.com/research/paper/68fa65d71e209131cea0cad4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64a71e209131cea05e51.md>)

[Implementation](<https://github.com/xingjunm/dimensionality-driven-learning>)

Recorded evaluation configuration: {"Architecture":"RN-44"}. This record does not assert independent reproduction.

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