The Prior-Predictive Value: A Paradigm of Nasty Multi-Dimensional Integrals

W. Von Der Linden, R. Preuss, V. Dose

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

At the heart of Bayesian model comparison lies the so-called prior-predictive value. In the important class of Quantified-MaxEnt applications analytic approximations are routinely used which often give rise to noise-fitting and ringing. We present an improved analytic expression which overcomes these shortcomings. In most interesting real-world problems, however, standard approximations and straight forward application of Markov-Chain Monte Carlo are hampered by the complicated structure of the likelihood in parameter space. At the Maxent workshop 1997 in Boise John Skilling suggested to employ a formalism, borrowed from Statistical physics, to compute the prior-predictive value. We have scrutinized his suggestion: IT WORKS!
Original languageEnglish
Title of host publicationMaximum Entropy and Bayesian Methods Garching, Germany 1998
EditorsWolfgang von der Linden, Volker Dose, Rainer Fischer, Roland Preuss
PublisherSpringer Netherlands
Pages319-326
Number of pages8
ISBN (Print)978-94-010-5982-4 978-94-011-4710-1
Publication statusPublished - 1999

Publication series

NameFundamental Theories of Physics
PublisherSpringer Netherlands

Keywords

  • Artificial Intelligence (incl. Robotics), Bayes factor, Coding and Information Theory, Discrete Mathematics in Computer Science, MCMC, model comparison, Prior-predictive value, Probability Theory and Stochastic Processes, Statistics, general

Fingerprint Dive into the research topics of 'The Prior-Predictive Value: A Paradigm of Nasty Multi-Dimensional Integrals'. Together they form a unique fingerprint.

Cite this