AI Paper Atlas

Research map / Adversarial robustness and security / Black-box and transfer attacks

A Classifier-Agnostic Zero-Shot Adversarial Attack Detection via CLIP

Hodaya Krakover, Meir Yossef Levi, Eyal Gofer, Guy Gilboa

ECCV 2026 · Adversarial robustness and security › Black-box and transfer attacks · approach: Robustness benchmarks and defenses · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use PCA aggregation instead of max percentile of prompt similarities (ablation)A4D · Tiny-ImageNet, DeiT-SmallAU-ROC (avg): 91.53 -> 92.47Improved
use PCA aggregation instead of min percentile (ablation)A4D · Tiny-ImageNet, DeiT-SmallAU-ROC (avg): 83.58 -> 92.47Improved
use PCA aggregation instead of mean percentile (ablation)A4D · Tiny-ImageNet, DeiT-SmallAU-ROC (avg): 91.04 -> 92.47Improved
use the CLIP prompt-similarity dictionary with PCA instead of naive distance to trajectory endpointA4D · Tiny-ImageNet, DeiT-Small, FGSMAU-ROC: 93.87 -> 99.24Improved
use prompt dictionary with PCA instead of naive endpoint distanceA4D · Tiny-ImageNet, DeiT-Small, CWAU-ROC: 66.31 -> 77.31Improved
use prompt dictionary with PCA instead of naive endpoint distanceA4D · Tiny-ImageNet, DeiT-Small, DeepFoolAU-ROC: 66.78 -> 78.62Improved
compare A4D with Wavelet noise baseline on Wide-ResNet averageA4D · Tiny-ImageNet, Wide-ResNetAU-ROC (avg): 78.36 -> 92.53Improved
compare A4D with Mahalanobis on ConvNeXt-Tiny averageA4D · Tiny-ImageNet, ConvNeXt-TinyAU-ROC (avg): 88.85 -> 84.97Hurt
compare A4D with attack-agnostic Mahalanobis* on ResNet34 averageA4D · Tiny-ImageNet, ResNet34AU-ROC (avg): 38.92 -> 93.32Improved
detect CW and DeepFool attacks whose perturbations follow image edgesA4D · Tiny-ImageNet, DeiT-SmallAU-ROC: 99.24 -> 77.31 (FGSM vs CW)Hurt
use generic prompt 'A noisy image' instead of curated dictionary (ablation)A4D · Tiny-ImageNetAU-ROCHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy