IDAL combines ResNet-50 with FPN and a new pseudo-label-weighted MMD loss to improve unsupervised domain adaptation accuracy on natural image benchmarks, but the gains are small and inconsistent across datasets.
Unsupervised domain adaptation by back- propagation
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
support 1representative citing papers
citing papers explorer
-
IDAL: Improved Domain Adaptive Learning for Natural Images Dataset
IDAL combines ResNet-50 with FPN and a new pseudo-label-weighted MMD loss to improve unsupervised domain adaptation accuracy on natural image benchmarks, but the gains are small and inconsistent across datasets.