|
Abstract:
|
The book "Computer Age Statistical Inference" follows developments in statistics from the early 1950's to the present, a period of enormous progress in statistical app- lications, but less so in basic inference. Empirical Bayes, initiated by Herbert Robbins in 1951, is the notable exception, offering, at least potentially, major advances in both theory and practice. This talk concerns the relation of empirical Bayes to more standard Bayesian ideas. To this end, an intermediate framework, "oracle Bayes", is introduced.
The goal here is to separate the empirical (ie frequentist) part of empirical Bayes from its Bayesian interpretations. The ideas are familiar ones from Robbins and his students, but modern developments, in theory but especially in computation, allow a clearer picture of what's at stake.
|
