# Image Classification on Tiny-ImageNet Symmetric Noise 0.6 (test)

Source: [Image Classification on Tiny-ImageNet Symmetric Noise 0.6 (test)](<https://www.sota2.com/research/sota/image-classification-on-tiny-imagenet-symmetric-noise-0-6-test>)

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

Data updated: 2026-02-26T05:34:10.600Z

Evaluation of Image Classification performance on the test split of Tiny ImageNet corrupted with 60% Symmetric Noise.

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) | Tiny-ImageNet Symmetric Noise 0.6 (test) |
| Metric | Accuracy (Best) |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 5 of 5 |
| 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 (Best)](<https://www.sota2.com/research/sota/image-classification-on-tiny-imagenet-symmetric-noise-0-6-test.md?metric=Accuracy%20(Best)>): Higher is better; unit: PERCENT.

- [Accuracy (Last)](<https://www.sota2.com/research/sota/image-classification-on-tiny-imagenet-symmetric-noise-0-6-test.md?metric=Accuracy%20(Last)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy (Best) | Result date | Report paper |
| --- | --- | --- | --- |
| DEFT | 80.6 | Not recorded | 68fa75001e209131cea6528a |
| ELR | 73.74 | Not recorded | 68fa75001e209131cea6528a |
| CE | 73.17 | Not recorded | 68fa75001e209131cea6528a |
| SCE | 71.76 | Not recorded | 68fa75001e209131cea6528a |
| GMM | 43.47 | Not recorded | 68fa75001e209131cea6528a |

### DEFT

Accuracy (Best): 80.6. Result ID: 69936e273a6c31484915dfd0.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75001e209131cea6528a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75001e209131cea6528a.md>)

[Implementation](<https://github.com/HotanLee/DeFT>)

Recorded evaluation configuration: {"Fine-tuning Protocol":"Ours","Backbone":"CLIP"}. This record does not assert independent reproduction.

### ELR

Accuracy (Best): 73.74. Result ID: 69936e273a6c31484915dfcd.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75001e209131cea6528a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa68a01e209131cea1d383.md>)

[Implementation](<https://github.com/shengliu66/ELR>)

Recorded evaluation configuration: {"Fine-tuning Protocol":"FFT","Backbone":"CLIP"}. This record does not assert independent reproduction.

### CE

Accuracy (Best): 73.17. Result ID: 69936e273a6c31484915dfcc.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75001e209131cea6528a.md>)

Recorded evaluation configuration: {"Fine-tuning Protocol":"FFT","Backbone":"CLIP"}. This record does not assert independent reproduction.

### SCE

Accuracy (Best): 71.76. Result ID: 69936e273a6c31484915dfce.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75001e209131cea6528a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa66351e209131cea0ec32.md>)

Recorded evaluation configuration: {"Fine-tuning Protocol":"FFT","Backbone":"CLIP"}. This record does not assert independent reproduction.

### GMM

Accuracy (Best): 43.47. Result ID: 69936e273a6c31484915dfcf.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75001e209131cea6528a.md>)

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

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

Recorded evaluation configuration: {"Fine-tuning Protocol":"FFT","Backbone":"CLIP"}. This record does not assert independent reproduction.

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