# Image Classification on CIFAR-10 (val) (Adversarial Robustness Metrics)

Source: [Image Classification on CIFAR-10 (val) (Adversarial Robustness Metrics)](<https://www.sota2.com/research/sota/image-classification-on-cifar-10-val-adversarial-robustness-metrics>)

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

Data updated: 2026-02-26T05:34:20.917Z

Evaluation of Image Classification performance on the CIFAR-10 validation set, specifically measuring robustness against various adversarial attacks (FGSM, PGD, MI-FGSM).

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 (val) |
| Metric | Accuracy (Natural) |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 16 of 16 |
| 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 (Natural)](<https://www.sota2.com/research/sota/image-classification-on-cifar-10-val-adversarial-robustness-metrics.md?metric=Accuracy%20(Natural)>): Higher is better; unit: PERCENT.

- [Accuracy (FGSM)](<https://www.sota2.com/research/sota/image-classification-on-cifar-10-val-adversarial-robustness-metrics.md?metric=Accuracy%20(FGSM)>): Higher is better; unit: PERCENT.

- [Accuracy (PGD-20)](<https://www.sota2.com/research/sota/image-classification-on-cifar-10-val-adversarial-robustness-metrics.md?metric=Accuracy%20(PGD-20)>): Higher is better; unit: PERCENT.

- [Accuracy (PGD-100)](<https://www.sota2.com/research/sota/image-classification-on-cifar-10-val-adversarial-robustness-metrics.md?metric=Accuracy%20(PGD-100)>): Higher is better; unit: PERCENT.

- [Accuracy (MI-FGSM)](<https://www.sota2.com/research/sota/image-classification-on-cifar-10-val-adversarial-robustness-metrics.md?metric=Accuracy%20(MI-FGSM)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy (Natural) | Result date | Report paper |
| --- | --- | --- | --- |
| Standard | 95.01 | Not recorded | 68fa6a8b1e209131cea28ac3 |
| FAT | 89.34 | Not recorded | 68fa6a8b1e209131cea28ac3 |
| TRADES-1 | 88.64 | Not recorded | 68fa6a8b1e209131cea28ac3 |
| PGD-7 | 87.25 | Not recorded | 68fa6a8b1e209131cea28ac3 |
| NADAR-A | 86.61 | Not recorded | 68fa6a8b1e209131cea28ac3 |
| NADAR-B | 86.23 | Not recorded | 68fa6a8b1e209131cea28ac3 |
| FreeAT-8 | 85.96 | Not recorded | 68fa6a8b1e209131cea28ac3 |
| TRADES-6 | 84.92 | Not recorded | 68fa6a8b1e209131cea28ac3 |
| PC-DARTS | 83.94 | Not recorded | 68fa6a8b1e209131cea28ac3 |
| RACL | 83.89 | Not recorded | 68fa6a8b1e209131cea28ac3 |
| DARTS | 83.75 | Not recorded | 68fa6a8b1e209131cea28ac3 |
| NASNet | 83.66 | Not recorded | 68fa6a8b1e209131cea28ac3 |
| AmoebaNet | 83.41 | Not recorded | 68fa6a8b1e209131cea28ac3 |
| RobNet-large | 78.57 | Not recorded | 68fa6a8b1e209131cea28ac3 |
| RobNet-medium | 78.33 | Not recorded | 68fa6a8b1e209131cea28ac3 |
| RobNet-small | 78.05 | Not recorded | 68fa6a8b1e209131cea28ac3 |

### Standard

Accuracy (Natural): 95.01. Result ID: 699ab6955698665ec4446f81.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6a8b1e209131cea28ac3.md>)

Recorded evaluation configuration: {"Category":"Standard","Backbone Params (M)":"46.2","Backbone FLOPs (G)":"6.7"}. This record does not assert independent reproduction.

### FAT

Accuracy (Natural): 89.34. Result ID: 699ab6955698665ec4446f83.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6a8b1e209131cea28ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa68461e209131cea1af64.md>)

Recorded evaluation configuration: {"Category":"Adversarial Training","Backbone Params (M)":"46.2","Backbone FLOPs (G)":"6.7"}. This record does not assert independent reproduction.

### TRADES-1

Accuracy (Natural): 88.64. Result ID: 699ab6955698665ec4446f85.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6a8b1e209131cea28ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65f71e209131cea0d897.md>)

Recorded evaluation configuration: {"Category":"Adversarial Training","Backbone Params (M)":"46.2","Backbone FLOPs (G)":"6.7"}. This record does not assert independent reproduction.

### PGD-7

Accuracy (Natural): 87.25. Result ID: 699ab6955698665ec4446f82.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6a8b1e209131cea28ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa665a1e209131cea0fbae.md>)

Recorded evaluation configuration: {"Category":"Adversarial Training","Backbone Params (M)":"46.2","Backbone FLOPs (G)":"6.7"}. This record does not assert independent reproduction.

### NADAR-A

Accuracy (Natural): 86.61. Result ID: 699ab6955698665ec4446f8f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6a8b1e209131cea28ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6a8b1e209131cea28ac3.md>)

Recorded evaluation configuration: {"Category":"Dilation","Backbone Params (M)":"46.2","Architecture Params (M)":"3.6","Backbone FLOPs (G)":"6.7","Architecture FLOPs (G)":"0.6"}. This record does not assert independent reproduction.

### NADAR-B

Accuracy (Natural): 86.23. Result ID: 699ab6955698665ec4446f90.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6a8b1e209131cea28ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6a8b1e209131cea28ac3.md>)

Recorded evaluation configuration: {"Category":"Dilation","Backbone Params (M)":"46.2","Architecture Params (M)":"4.4","Backbone FLOPs (G)":"6.7","Architecture FLOPs (G)":"0.7"}. This record does not assert independent reproduction.

### FreeAT-8

Accuracy (Natural): 85.96. Result ID: 699ab6955698665ec4446f84.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6a8b1e209131cea28ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa66ad1e209131cea11a3f.md>)

[Implementation](<https://github.com/ashafahi/free_adv_train>)

Recorded evaluation configuration: {"Category":"Adversarial Training","Backbone Params (M)":"46.2","Backbone FLOPs (G)":"6.7"}. This record does not assert independent reproduction.

### TRADES-6

Accuracy (Natural): 84.92. Result ID: 699ab6955698665ec4446f86.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6a8b1e209131cea28ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65f71e209131cea0d897.md>)

Recorded evaluation configuration: {"Category":"Adversarial Training","Backbone Params (M)":"46.2","Backbone FLOPs (G)":"6.7"}. This record does not assert independent reproduction.

### PC-DARTS

Accuracy (Natural): 83.94. Result ID: 699ab6955698665ec4446f8a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6a8b1e209131cea28ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa67511e209131cea1563d.md>)

[Implementation](<https://github.com/yuhuixu1993/PC-DARTS>)

Recorded evaluation configuration: {"Category":"Standard NAS","Architecture Params (M)":"3.6","Architecture FLOPs (G)":"0.6"}. This record does not assert independent reproduction.

### RACL

Accuracy (Natural): 83.89. Result ID: 699ab6955698665ec4446f8e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6a8b1e209131cea28ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6e741e209131cea3fedb.md>)

Recorded evaluation configuration: {"Category":"Robust NAS","Architecture Params (M)":"3.6","Architecture FLOPs (G)":"0.5"}. This record does not assert independent reproduction.

### DARTS

Accuracy (Natural): 83.75. Result ID: 699ab6955698665ec4446f89.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6a8b1e209131cea28ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65a91e209131cea0bb86.md>)

[Implementation](<https://github.com/quark0/darts>)

Recorded evaluation configuration: {"Category":"Standard NAS","Architecture Params (M)":"3.3","Architecture FLOPs (G)":"0.5"}. This record does not assert independent reproduction.

### NASNet

Accuracy (Natural): 83.66. Result ID: 699ab6955698665ec4446f88.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6a8b1e209131cea28ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64441e209131cea039e2.md>)

Recorded evaluation configuration: {"Category":"Standard NAS","Architecture Params (M)":"3.8","Architecture FLOPs (G)":"0.6"}. This record does not assert independent reproduction.

### AmoebaNet

Accuracy (Natural): 83.41. Result ID: 699ab6955698665ec4446f87.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6a8b1e209131cea28ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65691e209131cea0a262.md>)

[Implementation](<https://colab.research.google.com/github/google-research/google-research/blob/master/evolution/regularized_evolution_algorithm/regularized_evolution.ipynb>)

Recorded evaluation configuration: {"Category":"Standard NAS","Architecture Params (M)":"3.2","Architecture FLOPs (G)":"0.5"}. This record does not assert independent reproduction.

### RobNet-large

Accuracy (Natural): 78.57. Result ID: 699ab6955698665ec4446f8d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6a8b1e209131cea28ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa67411e209131cea154a0.md>)

Recorded evaluation configuration: {"Category":"Robust NAS","Architecture Params (M)":"6.9"}. This record does not assert independent reproduction.

### RobNet-medium

Accuracy (Natural): 78.33. Result ID: 699ab6955698665ec4446f8c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6a8b1e209131cea28ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa67411e209131cea154a0.md>)

Recorded evaluation configuration: {"Category":"Robust NAS","Architecture Params (M)":"5.7"}. This record does not assert independent reproduction.

### RobNet-small

Accuracy (Natural): 78.05. Result ID: 699ab6955698665ec4446f8b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6a8b1e209131cea28ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa67411e209131cea154a0.md>)

Recorded evaluation configuration: {"Category":"Robust NAS","Architecture Params (M)":"4.4"}. This record does not assert independent reproduction.

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