# Adversarial Robustness on CUB200 (Image Classification) Source: [Adversarial Robustness on CUB200 (Image Classification)]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:34:12.510Z Evaluation of adversarial robustness for Image Classification models on the CUB200 dataset, reporting standard accuracy alongside robustness metrics like PGD-k and Average Attack (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) | CUB200 | | Metric | Accuracy (Clean) | | Unit | PERCENT | | Direction | Higher 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 - [Accuracy (Clean)](): Higher is better; unit: PERCENT. - [Accuracy (PGD-10)](): Higher is better; unit: PERCENT. - [Accuracy (Average Attack)](): Higher is better; unit: PERCENT. ## Recorded results | Method | Accuracy (Clean) | Result date | Report paper | | --- | --- | --- | --- | | RoLI - Full-FT | 74.28 | Not recorded | 68fa72bf1e209131cea585bb | | RanLI - Full-FT | 70.83 | Not recorded | 68fa72bf1e209131cea585bb | | RanLI - Linear | 69.88 | Not recorded | 68fa72bf1e209131cea585bb | | RoLI - LoRA | 68.88 | Not recorded | 68fa72bf1e209131cea585bb | | RoLI - Bias | 68.14 | Not recorded | 68fa72bf1e209131cea585bb | | TWINS-AT | 68.09 | Not recorded | 68fa6ed11e209131cea4265b | | TWINS-AT | 68.09 | Not recorded | 68fa72bf1e209131cea585bb | | TWINS-AT | 67.64 | Not recorded | 68fa6ed11e209131cea4265b | | RanLI - Bias | 66.69 | Not recorded | 68fa72bf1e209131cea585bb | | AT | 65.74 | Not recorded | 68fa6ed11e209131cea4265b | | RanLI - LoRA | 64.74 | Not recorded | 68fa72bf1e209131cea585bb | | RoLI - Adapter | 64.2 | Not recorded | 68fa72bf1e209131cea585bb | | AutoLoRa | 62.78 | Not recorded | 68fa72bf1e209131cea585bb | | RanLI - VPT | 61.94 | Not recorded | 68fa72bf1e209131cea585bb | | TWINS-TRADES | 61.05 | Not recorded | 68fa6ed11e209131cea4265b | | TWINS-TRADES | 60.72 | Not recorded | 68fa6ed11e209131cea4265b | | RanLI - Adapter | 59.67 | Not recorded | 68fa72bf1e209131cea585bb | | TRADES | 58.92 | Not recorded | 68fa6ed11e209131cea4265b | ### RoLI - Full-FT Accuracy (Clean): 74.28. Result ID: 6992775a3a6c3148490f27b1. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"Swin Transformer","Initialization":"Robust Linear Initialization","Finetuning Protocol":"Full Finetuning"}. This record does not assert independent reproduction. ### RanLI - Full-FT Accuracy (Clean): 70.83. Result ID: 6992775a3a6c3148490f27ab. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Swin Transformer","Initialization":"Random Initialization","Finetuning Protocol":"Full Finetuning"}. This record does not assert independent reproduction. ### RanLI - Linear Accuracy (Clean): 69.88. Result ID: 6992775a3a6c3148490f27b0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Swin Transformer","Initialization":"Random Initialization","Finetuning Protocol":"Linear Probing"}. This record does not assert independent reproduction. ### RoLI - LoRA Accuracy (Clean): 68.88. Result ID: 6992775a3a6c3148490f27b3. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"Swin Transformer","Initialization":"Robust Linear Initialization","Finetuning Protocol":"LoRA"}. This record does not assert independent reproduction. ### RoLI - Bias Accuracy (Clean): 68.14. Result ID: 6992775a3a6c3148490f27b4. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"Swin Transformer","Initialization":"Robust Linear Initialization","Finetuning Protocol":"Bias-only"}. This record does not assert independent reproduction. ### TWINS-AT Accuracy (Clean): 68.09. Result ID: 699060623a6c314849001aff. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"warmup":"false"}. This record does not assert independent reproduction. ### TWINS-AT Accuracy (Clean): 68.09. Result ID: 6992775a3a6c3148490f27a9. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"ResNet-50"}. This record does not assert independent reproduction. ### TWINS-AT Accuracy (Clean): 67.64. Result ID: 699060623a6c314849001b00. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"warmup":"true"}. This record does not assert independent reproduction. ### RanLI - Bias Accuracy (Clean): 66.69. Result ID: 6992775a3a6c3148490f27ae. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"Swin Transformer","Initialization":"Random Initialization","Finetuning Protocol":"Bias-only"}. This record does not assert independent reproduction. ### AT Accuracy (Clean): 65.74. Result ID: 699060623a6c314849001afe. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### RanLI - LoRA Accuracy (Clean): 64.74. Result ID: 6992775a3a6c3148490f27ad. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"Swin Transformer","Initialization":"Random Initialization","Finetuning Protocol":"LoRA"}. This record does not assert independent reproduction. ### RoLI - Adapter Accuracy (Clean): 64.2. Result ID: 6992775a3a6c3148490f27b2. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"Swin Transformer","Initialization":"Robust Linear Initialization","Finetuning Protocol":"Adapter"}. This record does not assert independent reproduction. ### AutoLoRa Accuracy (Clean): 62.78. Result ID: 6992775a3a6c3148490f27aa. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"ResNet-50"}. This record does not assert independent reproduction. ### RanLI - VPT Accuracy (Clean): 61.94. Result ID: 6992775a3a6c3148490f27af. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"Swin Transformer","Initialization":"Random Initialization","Finetuning Protocol":"Visual Prompt Tuning"}. This record does not assert independent reproduction. ### TWINS-TRADES Accuracy (Clean): 61.05. Result ID: 699060623a6c314849001b03. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"warmup":"true"}. This record does not assert independent reproduction. ### TWINS-TRADES Accuracy (Clean): 60.72. Result ID: 699060623a6c314849001b02. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"warmup":"false"}. This record does not assert independent reproduction. ### RanLI - Adapter Accuracy (Clean): 59.67. Result ID: 6992775a3a6c3148490f27ac. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"Swin Transformer","Initialization":"Random Initialization","Finetuning Protocol":"Adapter"}. This record does not assert independent reproduction. ### TRADES Accuracy (Clean): 58.92. Result ID: 699060623a6c314849001b01. [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.