Computable priors sharpened into Occam's razors

Title: Computable priors sharpened into Occam's razors
Authors: Bickel, David R.
Date: 2016-12-30
Abstract: The posterior probabilities available under standard Bayesian statistics are computable, apply to small samples, and coherently incorporate previous information. Modifying their priors according to results from algorithmic information theory adds the advantage of implementing Occam's razor, giving simpler distributions of data higher prior probabilities.
CollectionMathématiques et statistiques // Mathematics and Statistics
algorithm161230.pdf115.45 kBAdobe PDFOpen