WBCP generalizes BQ-CP to weighted data via a weighted Dirichlet Dir(neff · w̃) that matches frequentist variance, extends stochastic dominance guarantees, and yields O(1/√neff) conditional coverage improvements, demonstrated on spatial prediction.
8): c=n eff is the unique variance-matching choice, bridging frequentist and Bayesian perspectives
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Weighted Bayesian Conformal Prediction
WBCP generalizes BQ-CP to weighted data via a weighted Dirichlet Dir(neff · w̃) that matches frequentist variance, extends stochastic dominance guarantees, and yields O(1/√neff) conditional coverage improvements, demonstrated on spatial prediction.