Empirical Bayes
A statistical estimation method that estimates the hyperparameters of a prior distribution from data. It lies between a completely unknown prior and a completely known prior, and can reduce estimation variance when data are limited.
Timeline (last 90 days)
The paper establishes the regret optimality of an NPMLE-based g-modeling procedure in Gaussian empirical Bayes problems with bounded priors.
SPLIT fits its formula via empirical Bayes and uses the fitted results to plan budgets on unseen tasks.
Empirical Bayes is a statistical estimation method that estimates prior hyperparameters from the data.
All Facts (3)
The paper establishes the regret optimality of an NPMLE-based g-modeling procedure in Gaussian empirical Bayes problems with bounded priors.
50%UnverifiedSPLIT fits its formula via empirical Bayes and uses the fitted results to plan budgets on unseen tasks.
50%UnverifiedEmpirical Bayes is a statistical estimation method that estimates prior hyperparameters from the data.
50%