Framework for federated learning with missing data that identifies conditions favoring complete-case estimators over inverse-probability weighting and proposes a calibrated weighting approach consistent if at least one candidate model is correct.
A general framework for imputation in surveys
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
stat.ME 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Federated Learning with Incomplete Data: When to Use Complete Cases and When to Weight
Framework for federated learning with missing data that identifies conditions favoring complete-case estimators over inverse-probability weighting and proposes a calibrated weighting approach consistent if at least one candidate model is correct.