# Classification on MNIST (Empirical Risk, Test Risk)

Source: [Classification on MNIST (Empirical Risk, Test Risk)](<https://www.sota2.com/research/sota/classification-on-mnist-empirical-risk-test-risk>)

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

Data updated: 2026-02-26T05:41:59.333Z

Evaluation of Classification performance on the MNIST dataset, reporting Empirical Risk (Rs(h)) and Test Risk (Rμ(h)).

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 | Classification |
| Dataset (including any stated split) | MNIST |
| Metric | Empirical Risk (Rs(h)) |
| Unit | SCALAR |
| Direction | Lower 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

- [Empirical Risk (Rs(h))](<https://www.sota2.com/research/sota/classification-on-mnist-empirical-risk-test-risk.md?metric=Empirical%20Risk%20(Rs(h))>): Lower is better; unit: SCALAR.

- [Test Risk (Rμ(h))](<https://www.sota2.com/research/sota/classification-on-mnist-empirical-risk-test-risk.md?metric=Test%20Risk%20(R%CE%BC(h))>): Lower is better; unit: SCALAR.

## Recorded results

| Method | Empirical Risk (Rs(h)) | Result date | Report paper |
| --- | --- | --- | --- |
| ERM | 0.065 | Not recorded | 68fa71091e209131cea4ecdd |
| Alg. 1 | 0.066 | Not recorded | 68fa71091e209131cea4ecdd |
| Alg. 1 | 0.084 | Not recorded | 68fa71091e209131cea4ecdd |
| Alg. 1 | 0.084 | Not recorded | 68fa71091e209131cea4ecdd |
| Alg. 1 | 0.112 | Not recorded | 68fa71091e209131cea4ecdd |
| Alg. 1 | 0.112 | Not recorded | 68fa71091e209131cea4ecdd |
| ERM | 0.119 | Not recorded | 68fa71091e209131cea4ecdd |
| ERM | 0.119 | Not recorded | 68fa71091e209131cea4ecdd |
| Alg. 1 | 0.2 | Not recorded | 68fa71091e209131cea4ecdd |
| Alg. 2 | 0.284 | Not recorded | 68fa71091e209131cea4ecdd |
| OGD | 0.378 | Not recorded | 68fa71091e209131cea4ecdd |

### ERM

Empirical Risk (Rs(h)): 0.065. Result ID: 699175963a6c314849084b34.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71091e209131cea4ecdd.md>)

Recorded evaluation configuration: {"Model Type":"Linear","Learning Setting":"Batch"}. This record does not assert independent reproduction.

### Alg. 1

Empirical Risk (Rs(h)): 0.066. Result ID: 699175963a6c314849084b33.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71091e209131cea4ecdd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71091e209131cea4ecdd.md>)

[Implementation](<https://github.com/paulviallard/NeurIPS23-PB-Wasserstein>)

Recorded evaluation configuration: {"Model Type":"Linear","Learning Setting":"Batch","epsilon":"1/√m","K":"0.2√m"}. This record does not assert independent reproduction.

### Alg. 1

Empirical Risk (Rs(h)): 0.084. Result ID: 699175963a6c314849084af7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71091e209131cea4ecdd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71091e209131cea4ecdd.md>)

[Implementation](<https://github.com/paulviallard/NeurIPS23-PB-Wasserstein>)

Recorded evaluation configuration: {"model":"Neural Network","epsilon (ε)":"1/√m","K":"0.2√m"}. This record does not assert independent reproduction.

### Alg. 1

Empirical Risk (Rs(h)): 0.084. Result ID: 699175963a6c314849084b57.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71091e209131cea4ecdd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71091e209131cea4ecdd.md>)

[Implementation](<https://github.com/paulviallard/NeurIPS23-PB-Wasserstein>)

Recorded evaluation configuration: {"Model":"Neural Network","Learning mode":"Batch","epsilon":"1/sqrt(m)"}. This record does not assert independent reproduction.

### Alg. 1

Empirical Risk (Rs(h)): 0.112. Result ID: 699175963a6c314849084af6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71091e209131cea4ecdd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71091e209131cea4ecdd.md>)

[Implementation](<https://github.com/paulviallard/NeurIPS23-PB-Wasserstein>)

Recorded evaluation configuration: {"model":"Neural Network","epsilon (ε)":"1/m","K":"0.2√m"}. This record does not assert independent reproduction.

### Alg. 1

Empirical Risk (Rs(h)): 0.112. Result ID: 699175963a6c314849084b56.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71091e209131cea4ecdd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71091e209131cea4ecdd.md>)

[Implementation](<https://github.com/paulviallard/NeurIPS23-PB-Wasserstein>)

Recorded evaluation configuration: {"Model":"Neural Network","Learning mode":"Batch","epsilon":"1/m"}. This record does not assert independent reproduction.

### ERM

Empirical Risk (Rs(h)): 0.119. Result ID: 699175963a6c314849084af8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71091e209131cea4ecdd.md>)

Recorded evaluation configuration: {"model":"Neural Network"}. This record does not assert independent reproduction.

### ERM

Empirical Risk (Rs(h)): 0.119. Result ID: 699175963a6c314849084b58.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71091e209131cea4ecdd.md>)

Recorded evaluation configuration: {"Model":"Neural Network","Learning mode":"Batch"}. This record does not assert independent reproduction.

### Alg. 1

Empirical Risk (Rs(h)): 0.2. Result ID: 699175963a6c314849084b32.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71091e209131cea4ecdd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71091e209131cea4ecdd.md>)

[Implementation](<https://github.com/paulviallard/NeurIPS23-PB-Wasserstein>)

Recorded evaluation configuration: {"Model Type":"Linear","Learning Setting":"Batch","epsilon":"1/m","K":"0.2√m"}. This record does not assert independent reproduction.

### Alg. 2

Empirical Risk (Rs(h)): 0.284. Result ID: 699175963a6c314849084b17.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71091e209131cea4ecdd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71091e209131cea4ecdd.md>)

[Implementation](<https://github.com/paulviallard/NeurIPS23-PB-Wasserstein>)

Recorded evaluation configuration: {"Model architecture":"Linear model","Learning protocol":"Online learning"}. This record does not assert independent reproduction.

### OGD

Empirical Risk (Rs(h)): 0.378. Result ID: 699175963a6c314849084b18.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71091e209131cea4ecdd.md>)

Recorded evaluation configuration: {"Model architecture":"Linear model","Learning protocol":"Online learning"}. This record does not assert independent reproduction.

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