AI Paper Atlas

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

Gradual Domain Adaptation without Indexed Intermediate Domains

Hong-You Chen, Wei-Lun Chao

NeurIPS 2021 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Source-free domain adaptation · 13 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 cycle-consistency refinement to progressive domain discriminator scores (ablation)IDOL · Rotated MNISTtarget accuracy: 85.7 -> 87.5Improved
add cycle-consistency refinement to progressive domain discriminator scores (ablation)IDOL · Portraitstarget accuracy: 83.4 -> 85.5Improved
add refinement to pre-defined index sequences (ablation)GDA · Rotated MNISTtarget accuracy: 87.9 -> 93.3Improved
add refinement to random sequences (ablation)GDA · Rotated MNISTtarget accuracy: 39.5 -> 57.5Improved
use progressive domain discriminator instead of classifier confidence as coarse score (ablation)IDOL · Rotated MNISTtarget accuracy: 45.5 -> 85.7Improved
make domain discriminator progressive (ablation)IDOL · Rotated MNISTtarget accuracy: 82.1 -> 85.7Improved
use IDOL sequence instead of random sequenceGDA · Rotated MNISTtarget accuracy: 39.5 -> 87.5Improved
use IDOL sequence instead of pre-defined yearsGDA · Portraitstarget accuracy: 85.8 -> 85.5No clear effect
add IDOL to outlier-domain sequenceGDA · Rotated MNIST outlier domainstarget accuracy: 77.0 -> 81.3Improved
use IDOL instead of confidence-based GDAGDA · CIFAR10-STLtarget accuracy: 77.1 -> 78.1Improved
use UDA self-training on target plus intermediate data vs source onlyUDA · CIFAR10-STLtarget accuracy: 76.6 -> 61.1Hurt

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