# Robust Image Classification on CIFAR-10 (test) (Symmetric & Asymmetric Noise) Source: [Robust Image Classification on CIFAR-10 (test) (Symmetric & Asymmetric Noise)]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:34:27.871Z Evaluation of Image Classification model robustness on the CIFAR-10 test set under varying levels of Symmetric and Asymmetric input noise corruption. 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) | CIFAR-10 (test) | | Metric | Accuracy (Symmetric Noise, η=0.2) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 25 of 25 | | 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 (Symmetric Noise, η=0.2)](): Higher is better; unit: PERCENT. - [Accuracy (Symmetric Noise, η=0.0)](): Higher is better; unit: PERCENT. - [Accuracy (Symmetric Noise, η=0.4)](): Higher is better; unit: PERCENT. - [Accuracy (Symmetric Noise, η=0.6)](): Higher is better; unit: PERCENT. - [Accuracy (Symmetric Noise, η=0.8)](): Higher is better; unit: PERCENT. - [Accuracy (Asymmetric Noise, η=0.2)](): Higher is better; unit: PERCENT. - [Accuracy (Asymmetric Noise, η=0.3)](): Higher is better; unit: PERCENT. - [Accuracy (Asymmetric Noise, η=0.4)](): Higher is better; unit: PERCENT. - [Accuracy (Symmetric Noise, η=0.5)](): Higher is better; unit: PERCENT. - [Accuracy (Symmetric Noise, η=0.9)](): Higher is better; unit: PERCENT. ## Recorded results | Method | Accuracy (Symmetric Noise, η=0.2) | Result date | Report paper | | --- | --- | --- | --- | | Ours (Modified Model) | 96.75 | Not recorded | 68fa6dad1e209131cea3b71c | | ContrastToDivide | 96.4 | Not recorded | 68fa6dad1e209131cea3b71c | | AugDesc | 96.3 | Not recorded | 68fa6dad1e209131cea3b71c | | PropMix | 96.09 | Not recorded | 68fa6dad1e209131cea3b71c | | Ours (Modified Model) | 95.99 | Not recorded | 68fa6dad1e209131cea3b71c | | ELR | 95.8 | Not recorded | 68fa6dad1e209131cea3b71c | | DivideMix | 95.7 | Not recorded | 68fa6dad1e209131cea3b71c | | Ours (Normal Model) | 95.47 | Not recorded | 68fa6dad1e209131cea3b71c | | Ours (Normal Model) | 95.04 | Not recorded | 68fa6dad1e209131cea3b71c | | FasTEN | 91.94 | Not recorded | 68fa6cf71e209131cea36e68 | | MSLC | 90.94 | Not recorded | 68fa6cf71e209131cea36e68 | | MLOC | 90.5 | Not recorded | 68fa6cf71e209131cea36e68 | | Deep kNN | 90.02 | Not recorded | 68fa6cf71e209131cea36e68 | | MW-Net | 89.76 | Not recorded | 68fa6cf71e209131cea36e68 | | MLaC | 89.75 | Not recorded | 68fa6cf71e209131cea36e68 | | GLC | 89.66 | Not recorded | 68fa6cf71e209131cea36e68 | | L2RW | 88.26 | Not recorded | 68fa6cf71e209131cea36e68 | | SL | 87.63 | Not recorded | 68fa66351e209131cea0ec32 | | Forward | 85.83 | Not recorded | 68fa66351e209131cea0ec32 | | GCE | 84.86 | Not recorded | 68fa66351e209131cea0ec32 | | Bootstrap | 83.95 | Not recorded | 68fa66351e209131cea0ec32 | | LSR | 83.49 | Not recorded | 68fa66351e209131cea0ec32 | | CE | 82.96 | Not recorded | 68fa66351e209131cea0ec32 | | Cross-Entropy | 82.7 | Not recorded | 68fa6dad1e209131cea3b71c | | D2L | 81.13 | Not recorded | 68fa66351e209131cea0ec32 | ### Ours (Modified Model) Accuracy (Symmetric Noise, η=0.2): 96.75. Result ID: 698fd47f3a6c314849fc2ab3. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Variant":"Modified","Test-Time Augmentation":true}. This record does not assert independent reproduction. ### ContrastToDivide Accuracy (Symmetric Noise, η=0.2): 96.4. Result ID: 698fd47f3a6c314849fc2aae. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### AugDesc Accuracy (Symmetric Noise, η=0.2): 96.3. Result ID: 698fd47f3a6c314849fc2aad. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### PropMix Accuracy (Symmetric Noise, η=0.2): 96.09. Result ID: 698fd47f3a6c314849fc2aaf. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Ours (Modified Model) Accuracy (Symmetric Noise, η=0.2): 95.99. Result ID: 698fd47f3a6c314849fc2ab2. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Variant":"Modified"}. This record does not assert independent reproduction. ### ELR Accuracy (Symmetric Noise, η=0.2): 95.8. Result ID: 698fd47f3a6c314849fc2aab. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### DivideMix Accuracy (Symmetric Noise, η=0.2): 95.7. Result ID: 698fd47f3a6c314849fc2aac. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Ours (Normal Model) Accuracy (Symmetric Noise, η=0.2): 95.47. Result ID: 698fd47f3a6c314849fc2ab1. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Variant":"Normal","Test-Time Augmentation":true}. This record does not assert independent reproduction. ### Ours (Normal Model) Accuracy (Symmetric Noise, η=0.2): 95.04. Result ID: 698fd47f3a6c314849fc2ab0. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Variant":"Normal"}. This record does not assert independent reproduction. ### FasTEN Accuracy (Symmetric Noise, η=0.2): 91.94. Result ID: 698f6f443a6c314849f88afe. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"ResNet-34"}. This record does not assert independent reproduction. ### MSLC Accuracy (Symmetric Noise, η=0.2): 90.94. Result ID: 698f6f443a6c314849f88afd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"ResNet-34"}. This record does not assert independent reproduction. ### MLOC Accuracy (Symmetric Noise, η=0.2): 90.5. Result ID: 698f6f443a6c314849f88afb. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"ResNet-34"}. This record does not assert independent reproduction. ### Deep kNN Accuracy (Symmetric Noise, η=0.2): 90.02. Result ID: 698f6f443a6c314849f88af9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"ResNet-34"}. This record does not assert independent reproduction. ### MW-Net Accuracy (Symmetric Noise, η=0.2): 89.76. Result ID: 698f6f443a6c314849f88af8. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"ResNet-34"}. This record does not assert independent reproduction. ### MLaC Accuracy (Symmetric Noise, η=0.2): 89.75. Result ID: 698f6f443a6c314849f88afc. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"ResNet-34"}. This record does not assert independent reproduction. ### GLC Accuracy (Symmetric Noise, η=0.2): 89.66. Result ID: 698f6f443a6c314849f88afa. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"ResNet-34"}. This record does not assert independent reproduction. ### L2RW Accuracy (Symmetric Noise, η=0.2): 88.26. Result ID: 698f6f443a6c314849f88af7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"ResNet-34"}. This record does not assert independent reproduction. ### SL Accuracy (Symmetric Noise, η=0.2): 87.63. Result ID: 698cdf9b8b517eb4331d54b3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Forward Accuracy (Symmetric Noise, η=0.2): 85.83. Result ID: 698cdf9b8b517eb4331d54b0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### GCE Accuracy (Symmetric Noise, η=0.2): 84.86. Result ID: 698cdf9b8b517eb4331d54b2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Bootstrap Accuracy (Symmetric Noise, η=0.2): 83.95. Result ID: 698cdf9b8b517eb4331d54af. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### LSR Accuracy (Symmetric Noise, η=0.2): 83.49. Result ID: 698cdf9b8b517eb4331d54ae. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### CE Accuracy (Symmetric Noise, η=0.2): 82.96. Result ID: 698cdf9b8b517eb4331d54ad. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Cross-Entropy Accuracy (Symmetric Noise, η=0.2): 82.7. Result ID: 698fd47f3a6c314849fc2aaa. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### D2L Accuracy (Symmetric Noise, η=0.2): 81.13. Result ID: 698cdf9b8b517eb4331d54b1. [Reporting paper]() [Method origin paper]() [Implementation]() 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.