# Targeted False Positive Inflation on ImageNet-1K (val)

Source: [Targeted False Positive Inflation on ImageNet-1K (val)](<https://www.sota2.com/research/sota/targeted-false-positive-inflation-on-imagenet-1k-val>)

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

Data updated: 2026-07-08T06:01:31.115Z

Evaluation of model robustness specifically measuring the inflation of False Positive Rates (FPR) when subjected to targeted perturbations or attacks on the ImageNet-1K validation set. Metrics focus on FPR lift and net added false positives.

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 | Targeted False Positive Inflation |
| Dataset (including any stated split) | ImageNet-1K (val) |
| Metric | Clean FPR |
| Unit | PERCENT |
| Direction | Lower is better |
| Results loaded | 3 of 3 |
| 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

- [Clean FPR](<https://www.sota2.com/research/sota/targeted-false-positive-inflation-on-imagenet-1k-val.md?metric=Clean%20FPR>): Lower is better; unit: PERCENT.

- [Perturbed FPR](<https://www.sota2.com/research/sota/targeted-false-positive-inflation-on-imagenet-1k-val.md?metric=Perturbed%20FPR>): Lower is better; unit: PERCENT.

- [FPR Lift](<https://www.sota2.com/research/sota/targeted-false-positive-inflation-on-imagenet-1k-val.md?metric=FPR%20Lift>): Lower is better; unit: SCALAR.

- [Net Added FPs](<https://www.sota2.com/research/sota/targeted-false-positive-inflation-on-imagenet-1k-val.md?metric=Net%20Added%20FPs>): Lower is better; unit: SCALAR.

- [Extra FPs per 1,000](<https://www.sota2.com/research/sota/targeted-false-positive-inflation-on-imagenet-1k-val.md?metric=Extra%20FPs%20per%201%2C000>): Lower is better; unit: SCALAR.

- [Corrected q \<= 0.05 Count](<https://www.sota2.com/research/sota/targeted-false-positive-inflation-on-imagenet-1k-val.md?metric=Corrected%20q%20%3C%3D%200.05%20Count>): Lower is better; unit: SCALAR.

## Recorded results

| Method | Clean FPR | Result date | Report paper |
| --- | --- | --- | --- |
| Statistical Adversaries (All proposed) | 5.005 | Not recorded | 6a4dd9a74da6617fe9198b83 |
| FFT-Hellinger | 5.005 | Not recorded | 6a4dd9a74da6617fe9198b83 |
| Bandpass-whitened | 5.005 | Not recorded | 6a4dd9a74da6617fe9198b83 |

### Statistical Adversaries (All proposed)

Clean FPR: 5.005. Result ID: 6a4de7ad60b88bb339d5e3c5.

[Reporting paper](<https://www.sota2.com/research/paper/6a4dd9a74da6617fe9198b83.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a4dd9a74da6617fe9198b83.md>)

Recorded evaluation configuration: {"Construction":"All proposed","Cells":"44","Evaluation Decisions":"439,560"}. This record does not assert independent reproduction.

### FFT-Hellinger

Clean FPR: 5.005. Result ID: 6a4de7ad60b88bb339d5e3c6.

[Reporting paper](<https://www.sota2.com/research/paper/6a4dd9a74da6617fe9198b83.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a4dd9a74da6617fe9198b83.md>)

Recorded evaluation configuration: {"Construction":"FFT-Hellinger","Cells":"24","Evaluation Decisions":"239,760"}. This record does not assert independent reproduction.

### Bandpass-whitened

Clean FPR: 5.005. Result ID: 6a4de7ad60b88bb339d5e3c7.

[Reporting paper](<https://www.sota2.com/research/paper/6a4dd9a74da6617fe9198b83.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a4dd9a74da6617fe9198b83.md>)

Recorded evaluation configuration: {"Construction":"Bandpass-whitened","Cells":"20","Evaluation Decisions":"199,800"}. This record does not assert independent reproduction.

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