AI Paper Atlas

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

Attract, Perturb, and Explore: Learning a Feature Alignment Network for Semi-supervised Domain Adaptation

Taekyung Kim, Changick Kim

ECCV 2020 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Source-free domain adaptation · 141 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 the attraction scheme to the supervised baseline (ablation)S+T · DomainNet (7 scenarios, 3-shot), AlexNetaccuracy (MEAN): 43.4 -> 46.1Improved
add the exploration scheme to attraction (ablation)Attraction scheme · DomainNet (7 scenarios, 3-shot), AlexNetaccuracy (MEAN): 46.1 -> 48.3Improved
add the perturbation scheme to attraction and exploration (ablation)Attract + Explore · DomainNet (7 scenarios, 3-shot), AlexNetaccuracy (MEAN): 48.3 -> 48.9Improved
use the proposed perturbation scheme instead of VAT (ablation)Perturbation · DomainNet (7 scenarios, 3-shot), AlexNetaccuracy (MEAN): 42.5 -> 47.9Improved
add DANN domain-adversarial alignment (ablation)S+T · DomainNet (7 scenarios, 3-shot), AlexNetaccuracy (MEAN): 43.4 -> 42.4Hurt
add MMD alignment (ablation)S+T · DomainNet (7 scenarios, 3-shot), AlexNetaccuracy (MEAN): 43.4 -> 43.8No clear effect
add DANN to the supervised baseline on Office-HomeS+T · Office-Home (12 scenarios), ResNet-34accuracy (MEAN): 66.2 -> 63.5Hurt
add CDAN to the supervised baseline on DomainNetS+T · DomainNet 3-shot, AlexNetaccuracy (MEAN): 43.4 -> 41.0Hurt
add ENT entropy minimization on ResNet-34 versus AlexNetS+T · DomainNet 3-shotaccuracy (MEAN): 60.0 -> 67.6 (ResNet-34); 43.4 -> 39.8 (AlexNet)Mixed
increase labeled target samples from 5 to 10 per classENT vs proposed · DomainNet, ResNet-34accuracy (MEAN): ENT 71.1 -> 74.0 vs MME 73.0; Ours 76.8Mixed

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