# Image Classification on TinyImageNet (test) (Adversarial Robustness) Source: [Image Classification on TinyImageNet (test) (Adversarial Robustness)]() Collection index: [llms.txt]() Data updated: 2026-06-02T09:08:09.371Z Evaluation of Image Classification performance on the test split of TinyImageNet, focusing on empirical adversarial robustness metrics including Clean accuracy, FGSM, PGD-10, and AA. 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 Classification | | Dataset (including any stated split) | TinyImageNet (test) | | Metric | Clean Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 12 of 12 | | 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 Accuracy](): Higher is better; unit: PERCENT. - [FGSM Accuracy](): Higher is better; unit: PERCENT. - [PGD-10 Accuracy](): Higher is better; unit: PERCENT. - [AA Accuracy](): Higher is better; unit: PERCENT. ## Recorded results | Method | Clean Accuracy | Result date | Report paper | | --- | --- | --- | --- | | Benign | 61.78 | Not recorded | 6a1e6385dbd7150506f8516d | | ELLE | 48.77 | Not recorded | 6a1e6385dbd7150506f8516d | | PGD-2 | 48.64 | Not recorded | 6a1e6385dbd7150506f8516d | | GradAlign | 47.61 | Not recorded | 6a1e6385dbd7150506f8516d | | FGSM-RS | 47.46 | Not recorded | 6a1e6385dbd7150506f8516d | | ZeroGrad | 46.08 | Not recorded | 6a1e6385dbd7150506f8516d | | MultiGrad | 45.42 | Not recorded | 6a1e6385dbd7150506f8516d | | SORA | 45.3 | Not recorded | 6a1e6385dbd7150506f8516d | | AAER | 44.72 | Not recorded | 6a1e6385dbd7150506f8516d | | N-FGSM | 44.66 | Not recorded | 6a1e6385dbd7150506f8516d | | FGSM | 36.61 | Not recorded | 6a1e6385dbd7150506f8516d | | NuAT | 34.8 | Not recorded | 6a1e6385dbd7150506f8516d | ### Benign Clean Accuracy: 61.78. Result ID: 6a1e9d69fbc373592c18f5bf. [Reporting paper]() Recorded evaluation configuration: {"Architecture":"PreActResNet-18"}. This record does not assert independent reproduction. ### ELLE Clean Accuracy: 48.77. Result ID: 6a1e9d69fbc373592c18f5c8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"PreActResNet-18"}. This record does not assert independent reproduction. ### PGD-2 Clean Accuracy: 48.64. Result ID: 6a1e9d69fbc373592c18f5ca. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"PreActResNet-18"}. This record does not assert independent reproduction. ### GradAlign Clean Accuracy: 47.61. Result ID: 6a1e9d69fbc373592c18f5c2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"PreActResNet-18"}. This record does not assert independent reproduction. ### FGSM-RS Clean Accuracy: 47.46. Result ID: 6a1e9d69fbc373592c18f5c1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"PreActResNet-18"}. This record does not assert independent reproduction. ### ZeroGrad Clean Accuracy: 46.08. Result ID: 6a1e9d69fbc373592c18f5c4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"PreActResNet-18"}. This record does not assert independent reproduction. ### MultiGrad Clean Accuracy: 45.42. Result ID: 6a1e9d69fbc373592c18f5c5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"PreActResNet-18"}. This record does not assert independent reproduction. ### SORA Clean Accuracy: 45.3. Result ID: 6a1e9d69fbc373592c18f5c9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"PreActResNet-18"}. This record does not assert independent reproduction. ### AAER Clean Accuracy: 44.72. Result ID: 6a1e9d69fbc373592c18f5c7. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Architecture":"PreActResNet-18"}. This record does not assert independent reproduction. ### N-FGSM Clean Accuracy: 44.66. Result ID: 6a1e9d69fbc373592c18f5c6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"PreActResNet-18"}. This record does not assert independent reproduction. ### FGSM Clean Accuracy: 36.61. Result ID: 6a1e9d69fbc373592c18f5c0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"PreActResNet-18"}. This record does not assert independent reproduction. ### NuAT Clean Accuracy: 34.8. Result ID: 6a1e9d69fbc373592c18f5c3. [Reporting paper]() Recorded evaluation configuration: {"Architecture":"PreActResNet-18"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.