AI Paper Atlas

Research map / Label-efficient and robust learning / Domain adaptation and generalization

Adversarial Robustness for Unsupervised Domain Adaptation

Muhammad Awais, Fengwei Zhou, Hang Xu, Lanqing Hong, Ping Luo, Sung-Ho Bae, Zhenguo Li

ICCV 2021 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Adversarial domain adaptation · 36 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
align UDA features with robust teachers (RFA) instead of normal MDDMDD · VisDA-2017PGD-20 robust accuracy (%): 4.03 -> 47.66Improved
add RFA instead of robust pre-training as backboneMDD · VisDA-2017clean accuracy (%): 67.72 -> 72.90Improved
replace normal backbone with adversarially pre-trained backboneCDAN · VisDA-2017clean accuracy (%): 74.23 -> 68.00Hurt
replace normal backbone with adversarially pre-trained backboneCDAN · VisDA-2017PGD-20 robust accuracy (%): 0 -> 41.67Improved
add RFA to CDANCDAN · VisDA-2017PGD-20 robust accuracy (%): 0 -> 43.49Improved
use weak adversarially pre-trained student instead of normal student (ablation)RFA · VisDA-2017PGD-20 robust accuracy (%): 7.63 -> 40.61Improved
use similarity preserving loss instead of L2 pairwise loss (ablation)RFA with MDD · VisDA-2017PGD-20 robust accuracy (%): 8.29 -> 47.66Improved
use multiple teachers instead of a single teacher (ablation)RFA · VisDA-2017clean accuracy (%): 70.31 -> 73.45Improved
use multiple teachers instead of a single teacher (ablation)RFA · VisDA-2017PGD-20 robust accuracy (%): 40.15 -> 40.87Improved
increase alpha from 100 to 1000 (ablation)RFA · VisDA-2017PGD-20 robust accuracy (%): 40.07 -> 47.66Improved
increase alpha from 500 to 5000 (ablation)RFA · VisDA-2017clean accuracy (%): 73.62 -> 70.31Hurt
apply RFA to Office-HomeMDD · Office-Homeclean accuracy (%): 67.91 -> 70.03Improved

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