# Image Reconstruction on Rotated MNIST SO(2) discretized (test)

Source: [Image Reconstruction on Rotated MNIST SO(2) discretized (test)](<https://www.sota2.com/research/sota/image-reconstruction-on-rotated-mnist-so-2-discretized-test>)

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

Data updated: 2026-02-26T06:01:45.895Z

Evaluation of image reconstruction performance on the test split of the Rotated MNIST SO(2) discretized dataset, typically measured using Reconstruction Loss.

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 Reconstruction |
| Dataset (including any stated split) | Rotated MNIST SO(2) discretized (test) |
| Metric | Reconstruction Loss |
| Unit | SCALAR |
| Direction | Lower is better |
| Results loaded | 6 of 6 |
| 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

- [Reconstruction Loss](<https://www.sota2.com/research/sota/image-reconstruction-on-rotated-mnist-so-2-discretized-test.md?metric=Reconstruction%20Loss>): Lower is better; unit: SCALAR.

## Recorded results

| Method | Reconstruction Loss | Result date | Report paper |
| --- | --- | --- | --- |
| GIAE | 0.0162 | Not recorded | 68fa72df1e209131cea5915f |
| Classical AE | 0.017 | Not recorded | 68fa72df1e209131cea5915f |
| Equivariant AE | 0.0189 | Not recorded | 68fa72df1e209131cea5915f |
| Equivariant AE | 0.0189 | Not recorded | 68fa72df1e209131cea5915f |
| Equivariant AE | 0.0189 | Not recorded | 68fa72df1e209131cea5915f |
| QAE | 0.0227 | Not recorded | 68fa72df1e209131cea5915f |

### GIAE

Reconstruction Loss: 0.0162. Result ID: 6992a17e3a6c314849104378.

[Reporting paper](<https://www.sota2.com/research/paper/68fa72df1e209131cea5915f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72df1e209131cea5915f.md>)

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

### Classical AE

Reconstruction Loss: 0.017. Result ID: 6992a17e3a6c314849104376.

[Reporting paper](<https://www.sota2.com/research/paper/68fa72df1e209131cea5915f.md>)

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

### Equivariant AE

Reconstruction Loss: 0.0189. Result ID: 6992a17e3a6c314849104379.

[Reporting paper](<https://www.sota2.com/research/paper/68fa72df1e209131cea5915f.md>)

Recorded evaluation configuration: {"Pooling Strategy":"norm"}. This record does not assert independent reproduction.

### Equivariant AE

Reconstruction Loss: 0.0189. Result ID: 6992a17e3a6c31484910437a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa72df1e209131cea5915f.md>)

Recorded evaluation configuration: {"Pooling Strategy":"angle"}. This record does not assert independent reproduction.

### Equivariant AE

Reconstruction Loss: 0.0189. Result ID: 6992a17e3a6c31484910437b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa72df1e209131cea5915f.md>)

Recorded evaluation configuration: {"Pooling Strategy":"complete"}. This record does not assert independent reproduction.

### QAE

Reconstruction Loss: 0.0227. Result ID: 6992a17e3a6c314849104377.

[Reporting paper](<https://www.sota2.com/research/paper/68fa72df1e209131cea5915f.md>)

Recorded evaluation configuration: {"Architecture":"quotient autoencoder"}. This record does not assert independent reproduction.

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