# Unsupervised Domain Adaptation on Digits (MNIST, USPS, SVHN)

Source: [Unsupervised Domain Adaptation on Digits (MNIST, USPS, SVHN)](<https://www.sota2.com/research/sota/unsupervised-domain-adaptation-on-digits-mnist-usps-svhn>)

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

Data updated: 2026-06-04T12:32:34.721Z

Evaluation of Unsupervised Domain Adaptation performance across the MNIST, USPS, and SVHN digit datasets, reporting average performance across specified source-target pairs (e.g., M-\>U, U-\>M, S-\>M).

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 | Unsupervised Domain Adaptation |
| Dataset (including any stated split) | Digits (MNIST, USPS, SVHN) |
| Metric | Accuracy (S-\>M) |
| 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 (S-\>M)](<https://www.sota2.com/research/sota/unsupervised-domain-adaptation-on-digits-mnist-usps-svhn.md?metric=Accuracy%20(S-%3EM)>): Higher is better; unit: PERCENT.

- [Accuracy (M-\>U)](<https://www.sota2.com/research/sota/unsupervised-domain-adaptation-on-digits-mnist-usps-svhn.md?metric=Accuracy%20(M-%3EU)>): Higher is better; unit: PERCENT.

- [Accuracy (U-\>M)](<https://www.sota2.com/research/sota/unsupervised-domain-adaptation-on-digits-mnist-usps-svhn.md?metric=Accuracy%20(U-%3EM)>): Higher is better; unit: PERCENT.

- [Average Accuracy](<https://www.sota2.com/research/sota/unsupervised-domain-adaptation-on-digits-mnist-usps-svhn.md?metric=Average%20Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy (S-\>M) | Result date | Report paper |
| --- | --- | --- | --- |
| Target Risk (Upper Bound) | 93.51 | Not recorded | 6a2106ffc10484bac23f0781 |
| DEV | 93.18 | Not recorded | 6a2106ffc10484bac23f0781 |
| Original | 92.4 | Not recorded | 6a2106ffc10484bac23f0781 |
| ILA-DA (with CDAN) | 92.3 | Not recorded | 68fa69b41e209131cea239a7 |
| ILA-DA (with DANN) | 91.84 | Not recorded | 68fa69b41e209131cea239a7 |
| UNIT | 90.5 | Not recorded | 68fa65391e209131cea09230 |
| CyCADA | 90.4 | Not recorded | 68fa65391e209131cea09230 |
| CDAN+E | 89.2 | Not recorded | 68fa65391e209131cea09230 |
| CDAN | 88.5 | Not recorded | 68fa65391e209131cea09230 |
| CDAN | 88.5 | Not recorded | 68fa69b41e209131cea239a7 |
| ATT | 85 | Not recorded | 68fa69b41e209131cea239a7 |
| DANN | 83.11 | Not recorded | 68fa69b41e209131cea239a7 |
| DSN | 82.7 | Not recorded | 68fa69b41e209131cea239a7 |
| Source Risk | 77.58 | Not recorded | 6a2106ffc10484bac23f0781 |
| ADDA | 76 | Not recorded | 68fa69b41e209131cea239a7 |
| Source Only | 67.1 | Not recorded | 68fa69b41e209131cea239a7 |

### Target Risk (Upper Bound)

Accuracy (S-\>M): 93.51. Result ID: 6a217047fbc373592c1e987f.

[Reporting paper](<https://www.sota2.com/research/paper/6a2106ffc10484bac23f0781.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63e81e209131cea015de.md>)

Recorded evaluation configuration: {"Base Model":"GTA","Validation Method":"Target Risk (Upper Bound)"}. This record does not assert independent reproduction.

### DEV

Accuracy (S-\>M): 93.18. Result ID: 6a217047fbc373592c1e987e.

[Reporting paper](<https://www.sota2.com/research/paper/6a2106ffc10484bac23f0781.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a2106ffc10484bac23f0781.md>)

Recorded evaluation configuration: {"Base Model":"GTA","Validation Method":"DEV"}. This record does not assert independent reproduction.

### Original

Accuracy (S-\>M): 92.4. Result ID: 6a217047fbc373592c1e987c.

[Reporting paper](<https://www.sota2.com/research/paper/6a2106ffc10484bac23f0781.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64481e209131cea03ae2.md>)

[Implementation](<https://goo.gl/zUVeqC>)

Recorded evaluation configuration: {"Base Model":"GTA","Validation Method":"Original"}. This record does not assert independent reproduction.

### ILA-DA (with CDAN)

Accuracy (S-\>M): 92.3. Result ID: 698dda50ffc3901de0991944.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69b41e209131cea239a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69b41e209131cea239a7.md>)

[Implementation](<https://github.com/astuti/ILA-DA>)

Recorded evaluation configuration: {"Backbone":"LeNet","Base Method":"CDAN","k":3,"mu":0.75}. This record does not assert independent reproduction.

### ILA-DA (with DANN)

Accuracy (S-\>M): 91.84. Result ID: 698dda50ffc3901de0991942.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69b41e209131cea239a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69b41e209131cea239a7.md>)

[Implementation](<https://github.com/astuti/ILA-DA>)

Recorded evaluation configuration: {"Backbone":"LeNet","Base Method":"DANN","k":3,"mu":0.75}. This record does not assert independent reproduction.

### UNIT

Accuracy (S-\>M): 90.5. Result ID: 698ca33c0b614d12cff222f5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa65391e209131cea09230.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64991e209131cea05654.md>)

[Implementation](<https://github.com/mingyuliutw/unit>)

Recorded evaluation configuration: {"Backbone":"ResNet-50"}. This record does not assert independent reproduction.

### CyCADA

Accuracy (S-\>M): 90.4. Result ID: 698ca33c0b614d12cff222f6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa65391e209131cea09230.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63e81e209131cea015de.md>)

Recorded evaluation configuration: {"Backbone":"ResNet-50"}. This record does not assert independent reproduction.

### CDAN+E

Accuracy (S-\>M): 89.2. Result ID: 698ca33c0b614d12cff222f8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa65391e209131cea09230.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65391e209131cea09230.md>)

[Implementation](<http://github.com/thuml/CDAN>)

Recorded evaluation configuration: {"Backbone":"ResNet-50"}. This record does not assert independent reproduction.

### CDAN

Accuracy (S-\>M): 88.5. Result ID: 698ca33c0b614d12cff222f7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa65391e209131cea09230.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65391e209131cea09230.md>)

[Implementation](<http://github.com/thuml/CDAN>)

Recorded evaluation configuration: {"Backbone":"ResNet-50"}. This record does not assert independent reproduction.

### CDAN

Accuracy (S-\>M): 88.5. Result ID: 698dda50ffc3901de0991943.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69b41e209131cea239a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65391e209131cea09230.md>)

[Implementation](<http://github.com/thuml/CDAN>)

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

### ATT

Accuracy (S-\>M): 85. Result ID: 698dda50ffc3901de0991941.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69b41e209131cea239a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63541e209131ce9fe3d3.md>)

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

### DANN

Accuracy (S-\>M): 83.11. Result ID: 698dda50ffc3901de099193e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69b41e209131cea239a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa629d1e209131ce9fa35a.md>)

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

### DSN

Accuracy (S-\>M): 82.7. Result ID: 698dda50ffc3901de0991940.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69b41e209131cea239a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa62b91e209131ce9fade7.md>)

[Implementation](<https://github.com/tensorflow/models/>)

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

### Source Risk

Accuracy (S-\>M): 77.58. Result ID: 6a217047fbc373592c1e987d.

[Reporting paper](<https://www.sota2.com/research/paper/6a2106ffc10484bac23f0781.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa627d1e209131ce9f988d.md>)

Recorded evaluation configuration: {"Base Model":"GTA","Validation Method":"Source Risk"}. This record does not assert independent reproduction.

### ADDA

Accuracy (S-\>M): 76. Result ID: 698dda50ffc3901de099193f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69b41e209131cea239a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa66d21e209131cea129e2.md>)

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

### Source Only

Accuracy (S-\>M): 67.1. Result ID: 698dda50ffc3901de099193d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69b41e209131cea239a7.md>)

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

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