AI Paper Atlas

Research map / Adversarial robustness and security / Certified and empirical robustness

DRIFT: Divergent Response in Filtered Transformations for Robust Adversarial Defense

Amira Guesmi, Muhammad Shafique

ICLR 2026 · Adversarial robustness and security › Certified and empirical robustness · approach: Attack evaluation and defense critique · 1 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
add Jacobian separation loss (ablation)CE + adversarial training · ImageNet adaptive PGD (ResNet-v2)robust accuracy: 3.70 -> 39.80Improved
add logit-VJP separation loss (ablation)CE + adversarial training · ImageNet adaptive PGD (ResNet-v2)robust accuracy: 3.70 -> 47.61Improved
combine JS and LVJP losses (ablation)CE + LVJP + adv · ImageNet adaptive PGD (ResNet-v2)robust accuracy: 47.61 -> 53.78Improved
combine JS and LVJP losses (ablation)CE + LVJP + adv · ImageNet adaptive PGD (ViT-B/16)robust accuracy: 49.73 -> 56.74Improved
separation losses vs clean accuracy (ablation)DRIFT · ImageNet non-adaptiveclean accuracy: 75.66 -> 76.19 (ResNet-v2)No clear effect
use residual two-conv filter instead of single conv (ablation)DRIFT filter · ImageNet adaptive PGD (ViT-B/16)robust accuracy: 42.5 -> 56.7Improved
use residual two-conv filter instead of deeper four-conv filter (ablation)DRIFT filter · ImageNet adaptive PGD (ViT-B/16)robust accuracy: 47.9 -> 56.7Improved
allow attacker BPDA in addition to EOTDRIFT · ImageNet adaptive PGD (ResNet-v2)robust accuracy: 53.78 -> 60.19Mixed
learned filter ensemble vs diffusion purificationDiffPure · ImageNet AutoAttack non-adaptive (ResNet-v2)robust accuracy: 67.01 -> 74.30Improved
increase EOT samples K from 5 to 20DRIFT · ImageNet PGD l_inf eps 4/255 (ViT-B/16)robust accuracy: 57.62 -> 38.57Hurt
replace diffusion purification with learned filtersDiffPure · ResNet-50 latencylatency (s): 17.07 -> 0.0004Improved

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