# Model Extraction Attack Robustness on TinyImageNet

Source: [Model Extraction Attack Robustness on TinyImageNet](<https://www.sota2.com/research/sota/model-extraction-attack-robustness-on-tinyimagenet>)

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

Data updated: 2026-02-26T05:31:19.648Z

Evaluation of model robustness specifically against Model Extraction Attacks targeting models trained on the TinyImageNet dataset, reporting Accuracy (Acc) and Win/Success Rate (WSR).

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 | Model Extraction Attack Robustness |
| Dataset (including any stated split) | TinyImageNet |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 7 of 7 |
| 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](<https://www.sota2.com/research/sota/model-extraction-attack-robustness-on-tinyimagenet.md?metric=Accuracy>): Higher is better; unit: PERCENT.

- [Win/Success Rate](<https://www.sota2.com/research/sota/model-extraction-attack-robustness-on-tinyimagenet.md?metric=Win%2FSuccess%20Rate>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| ComMark | 0.3914 | Not recorded | 69438aa4423ea422657295c8 |
| Content | 0.3702 | Not recorded | 69438aa4423ea422657295c8 |
| MAB | 0.3697 | Not recorded | 69438aa4423ea422657295c8 |
| Noise | 0.3679 | Not recorded | 69438aa4423ea422657295c8 |
| Abstract | 0.232 | Not recorded | 69438aa4423ea422657295c8 |
| BlindMark | 0.2094 | Not recorded | 69438aa4423ea422657295c8 |
| MEAD | 0.1716 | Not recorded | 69438aa4423ea422657295c8 |

### ComMark

Accuracy: 0.3914. Result ID: 699701515698665ec44e5a30.

[Reporting paper](<https://www.sota2.com/research/paper/69438aa4423ea422657295c8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69438aa4423ea422657295c8.md>)

Recorded evaluation configuration: {"Victim Model Architecture":"ResNet18","Victim Training Dataset":"CIFAR10","Adversary Model Architecture":"DenseNet161","Adversary Query Set":"TinyImageNet"}. This record does not assert independent reproduction.

### Content

Accuracy: 0.3702. Result ID: 699701515698665ec44e5a2b.

[Reporting paper](<https://www.sota2.com/research/paper/69438aa4423ea422657295c8.md>)

Recorded evaluation configuration: {"Victim Model Architecture":"ResNet18","Victim Training Dataset":"CIFAR10","Adversary Model Architecture":"DenseNet161","Adversary Query Set":"TinyImageNet"}. This record does not assert independent reproduction.

### MAB

Accuracy: 0.3697. Result ID: 699701515698665ec44e5a2f.

[Reporting paper](<https://www.sota2.com/research/paper/69438aa4423ea422657295c8.md>)

Recorded evaluation configuration: {"Victim Model Architecture":"ResNet18","Victim Training Dataset":"CIFAR10","Adversary Model Architecture":"DenseNet161","Adversary Query Set":"TinyImageNet"}. This record does not assert independent reproduction.

### Noise

Accuracy: 0.3679. Result ID: 699701515698665ec44e5a2d.

[Reporting paper](<https://www.sota2.com/research/paper/69438aa4423ea422657295c8.md>)

Recorded evaluation configuration: {"Victim Model Architecture":"ResNet18","Victim Training Dataset":"CIFAR10","Adversary Model Architecture":"DenseNet161","Adversary Query Set":"TinyImageNet"}. This record does not assert independent reproduction.

### Abstract

Accuracy: 0.232. Result ID: 699701515698665ec44e5a2a.

[Reporting paper](<https://www.sota2.com/research/paper/69438aa4423ea422657295c8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64821e209131cea05086.md>)

Recorded evaluation configuration: {"Victim Model Architecture":"ResNet18","Victim Training Dataset":"CIFAR10","Adversary Model Architecture":"DenseNet161","Adversary Query Set":"TinyImageNet"}. This record does not assert independent reproduction.

### BlindMark

Accuracy: 0.2094. Result ID: 699701515698665ec44e5a2e.

[Reporting paper](<https://www.sota2.com/research/paper/69438aa4423ea422657295c8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa66981e209131cea11339.md>)

Recorded evaluation configuration: {"Victim Model Architecture":"ResNet18","Victim Training Dataset":"CIFAR10","Adversary Model Architecture":"DenseNet161","Adversary Query Set":"TinyImageNet"}. This record does not assert independent reproduction.

### MEAD

Accuracy: 0.1716. Result ID: 699701515698665ec44e5a2c.

[Reporting paper](<https://www.sota2.com/research/paper/69438aa4423ea422657295c8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72001e209131cea540dd.md>)

Recorded evaluation configuration: {"Victim Model Architecture":"ResNet18","Victim Training Dataset":"CIFAR10","Adversary Model Architecture":"DenseNet161","Adversary Query Set":"TinyImageNet"}. This record does not assert independent reproduction.

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