# Trojan Defense on ImageNet subset (200 classes) (test)

Source: [Trojan Defense on ImageNet subset (200 classes) (test)](<https://www.sota2.com/research/sota/trojan-defense-on-imagenet-subset-200-classes-test>)

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

Data updated: 2026-02-26T05:37:27.751Z

Evaluation of Trojan Defense mechanisms on a 200-class subset of ImageNet (test split), reporting Backdoor Accuracy (BA), Attack Success Rate (ASR), and computational Runtime.

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 | Trojan Defense |
| Dataset (including any stated split) | ImageNet subset (200 classes) (test) |
| Metric | Backdoor Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 2 of 2 |
| 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

- [Backdoor Accuracy](<https://www.sota2.com/research/sota/trojan-defense-on-imagenet-subset-200-classes-test.md?metric=Backdoor%20Accuracy>): Higher is better; unit: PERCENT.

- [Attack Success Rate](<https://www.sota2.com/research/sota/trojan-defense-on-imagenet-subset-200-classes-test.md?metric=Attack%20Success%20Rate>): Higher is better; unit: PERCENT.

- [Runtime (h)](<https://www.sota2.com/research/sota/trojan-defense-on-imagenet-subset-200-classes-test.md?metric=Runtime%20(h)>): Lower is better; unit: TIME.

## Recorded results

| Method | Backdoor Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Native Training | 85.12 | Not recorded | 68fa6dc51e209131cea3c033 |
| NONE | 83.46 | Not recorded | 68fa6dc51e209131cea3c033 |

### Native Training

Backdoor Accuracy: 85.12. Result ID: 698fdccd3a6c314849fc63bd.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6dc51e209131cea3c033.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### NONE

Backdoor Accuracy: 83.46. Result ID: 698fdccd3a6c314849fc63be.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6dc51e209131cea3c033.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6dc51e209131cea3c033.md>)

[Implementation](<https://github.com/RU-System-Software-and-Security/NONE>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

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