# Certified Robustness on ImageNet (L2 Perturbation Bounds) Source: [Certified Robustness on ImageNet (L2 Perturbation Bounds)]() Collection index: [llms.txt]() Data updated: 2026-05-26T10:49:47.197Z Evaluation of model certified robustness on the standard ImageNet dataset, reporting certified accuracy across various L2 perturbation radii (0.0 to 1.0). 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 | Certified Robustness | | Dataset (including any stated split) | ImageNet | | Metric | Inference Time (s) | | Unit | TIME | | Direction | Lower is better | | Results loaded | 18 of 18 | | 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 - [Inference Time (s)](): Lower is better; unit: TIME. - [Certified Accuracy (r=0.0)](): Higher is better; unit: PERCENT. - [Certified Accuracy (r=0.12)](): Higher is better; unit: PERCENT. - [Certified Accuracy (r=0.25)](): Higher is better; unit: PERCENT. - [Certified Accuracy (r=0.5)](): Higher is better; unit: PERCENT. - [Certified Accuracy (r=1.0)](): Higher is better; unit: PERCENT. ## Recorded results | Method | Inference Time (s) | Result date | Report paper | | --- | --- | --- | --- | | AS | 6.2 | Not recorded | 69f1907655a3c32af89fb9b2 | | AS | 6.2 | Not recorded | 69f1907655a3c32af89fb9b2 | | LBS | 10.8 | Not recorded | 69f1907655a3c32af89fb9b2 | | Betting CS | 24.9 | Not recorded | 69f1907655a3c32af89fb9b2 | | Betting CS | 24.9 | Not recorded | 69f1907655a3c32af89fb9b2 | | ISS | 109.9 | Not recorded | 69f1907655a3c32af89fb9b2 | | ISS | 109.9 | Not recorded | 69f1907655a3c32af89fb9b2 | | Ensemble | 120.3 | Not recorded | 69f1907655a3c32af89fb9b2 | | UCAN | 121.6 | Not recorded | 69f1907655a3c32af89fb9b2 | | UCAN | 121.6 | Not recorded | 69f1907655a3c32af89fb9b2 | | ADRE | 122.7 | Not recorded | 69f1907655a3c32af89fb9b2 | | ADRE | 122.8 | Not recorded | 69f1907655a3c32af89fb9b2 | | RS | 123.5 | Not recorded | 69f1907655a3c32af89fb9b2 | | RS | 123.5 | Not recorded | 69f1907655a3c32af89fb9b2 | | GENELLIPS | 163.74 | Not recorded | 6a152965dbd7150506f7e325 | | Randomized Smoothing | 180.9 | Not recorded | 6a152965dbd7150506f7e325 | | Denoising with Pre-trained Diffusion Models | 1535.94 | Not recorded | 6a152965dbd7150506f7e325 | | DensePure | 2188.2 | Not recorded | 6a152965dbd7150506f7e325 | ### AS Inference Time (s): 6.2. Result ID: 69f1b0d047f34ce924d7bfe6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training Augmentation":"standard (w/o)","sigma":"0.5","N":"100,000"}. This record does not assert independent reproduction. ### AS Inference Time (s): 6.2. Result ID: 69f1b0d047f34ce924d7bfe9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training Augmentation":"noise-augmented (w/)","sigma":"0.5","N":"100,000"}. This record does not assert independent reproduction. ### LBS Inference Time (s): 10.8. Result ID: 69f1b0d047f34ce924d7bfec. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"sigma":"0.5","N":"100,000"}. This record does not assert independent reproduction. ### Betting CS Inference Time (s): 24.9. Result ID: 69f1b0d047f34ce924d7bfe5. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Training Augmentation":"standard (w/o)","sigma":"0.5","N":"100,000"}. This record does not assert independent reproduction. ### Betting CS Inference Time (s): 24.9. Result ID: 69f1b0d047f34ce924d7bfe8. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Training Augmentation":"noise-augmented (w/)","sigma":"0.5","N":"100,000"}. This record does not assert independent reproduction. ### ISS Inference Time (s): 109.9. Result ID: 69f1b0d047f34ce924d7bfe1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training Augmentation":"standard (w/o)","sigma":"0.5","N":"100,000"}. This record does not assert independent reproduction. ### ISS Inference Time (s): 109.9. Result ID: 69f1b0d047f34ce924d7bfe4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training Augmentation":"noise-augmented (w/)","sigma":"0.5","N":"100,000"}. This record does not assert independent reproduction. ### Ensemble Inference Time (s): 120.3. Result ID: 69f1b0d047f34ce924d7bfea. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training Augmentation":"noise-augmented (w/)","sigma":"0.5","N":"100,000"}. This record does not assert independent reproduction. ### UCAN Inference Time (s): 121.6. Result ID: 69f1b0d047f34ce924d7bfe7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training Augmentation":"standard (w/o)","sigma":"0.5","N":"100,000"}. This record does not assert independent reproduction. ### UCAN Inference Time (s): 121.6. Result ID: 69f1b0d047f34ce924d7bfeb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training Augmentation":"noise-augmented (w/)","sigma":"0.5","N":"100,000"}. This record does not assert independent reproduction. ### ADRE Inference Time (s): 122.7. Result ID: 69f1b0d047f34ce924d7bfe0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training Augmentation":"standard (w/o)","sigma":"0.5","N":"100,000"}. This record does not assert independent reproduction. ### ADRE Inference Time (s): 122.8. Result ID: 69f1b0d047f34ce924d7bfe3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training Augmentation":"noise-augmented (w/)","sigma":"0.5","N":"100,000"}. This record does not assert independent reproduction. ### RS Inference Time (s): 123.5. Result ID: 69f1b0d047f34ce924d7bfdf. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training Augmentation":"standard (w/o)","sigma":"0.5","N":"100,000"}. This record does not assert independent reproduction. ### RS Inference Time (s): 123.5. Result ID: 69f1b0d047f34ce924d7bfe2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training Augmentation":"noise-augmented (w/)","sigma":"0.5","N":"100,000"}. This record does not assert independent reproduction. ### GENELLIPS Inference Time (s): 163.74. Result ID: 6a157ac9fbc373592c0dae70. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Randomized Smoothing Inference Time (s): 180.9. Result ID: 6a157ac9fbc373592c0dae6f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Denoising with Pre-trained Diffusion Models Inference Time (s): 1535.94. Result ID: 6a157ac9fbc373592c0dae6e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### DensePure Inference Time (s): 2188.2. Result ID: 6a157ac9fbc373592c0dae6d. [Reporting paper]() [Method origin paper]() 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.