Derives target-domain prediction models and proposes distribution matching to estimate subpopulation proportions in unsupervised domain adaptation with an unobservable source subpopulation, with asymptotic guarantees and an upper bound on prediction error.
This inequality is analogous to Hoeffding’s inequality and provides a non-asymptotic concentration bound on the estimation error
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
stat.ML 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Unsupervised Domain Adaptation for Binary Classification with an Unobservable Source Subpopulation
Derives target-domain prediction models and proposes distribution matching to estimate subpopulation proportions in unsupervised domain adaptation with an unobservable source subpopulation, with asymptotic guarantees and an upper bound on prediction error.