AI Paper Atlas

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

Semi-Supervised Domain Adaptation via Minimax Entropy

Kuniaki Saito, Donghyun Kim, Stan Sclaroff, Trevor Darrell, Kate Saenko

ICCV 2019 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Adversarial domain adaptation · 722 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 entropy maximization on the classifier (minimax) instead of plain entropy minimization (ablation)ENT (ResNet34) · DomainNet mean 1-shotaccuracy: 62.6 -> 66.4Improved
add entropy maximization on the classifier (minimax) instead of plain entropy minimization (ablation)ENT (AlexNet) · DomainNet mean 1-shotaccuracy: 29.1 -> 44.2Improved
use the l2-normalized cosine classifier with temperature instead of a standard linear layer (ablation)MME (AlexNet) · DomainNet R to C 1-shotaccuracy: 44.9 -> 48.9Improved
use the l2-normalized cosine classifier with temperature instead of a standard linear layer (ablation)S+T (AlexNet) · DomainNet R to C 1-shotaccuracy: 41.4 -> 43.3Improved
use separate BN statistics for labeled and unlabeled batches (ablation)ENT (ResNet34) · DomainNetaccuracy: 63.6 -> 68.9Improved
use separate BN statistics for labeled and unlabeled batches (ablation)MME (ResNet34) · DomainNetaccuracy: 69.5 -> 69.6No clear effect
add virtual adversarial training loss on unlabeled targetS+T (AlexNet) · DomainNet R to C 3-shotaccuracy: 47.1 -> 46.1Hurt
add MME to S+TS+T (ResNet34) · DomainNet mean 1-shotaccuracy: 56.9 -> 66.4Improved
add entropy minimization (ENT) to S+TS+T (AlexNet) · DomainNet mean 1-shotaccuracy: 40.0 -> 29.1Hurt
add DANN domain classifier to S+TS+T (VGG16) · DomainNet mean 1-shotaccuracy: 50.5 -> 45.4Hurt
add MME in unsupervised DA settingSource-only (AlexNet) · DomainNet meanaccuracy: 37.2 -> 40.2Improved
add entropy minimization (ENT) to S+T with more labelsS+T · DomainNetaccuracy: 5 labeled examples per classMixed

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