Evidential model ranking without likelihoods

Vladislav Vyshemirsky

Research output: Contribution to conferenceSpeech

Abstract

We present a probabilistic formulation of the Approximate Bayesian Computation scheme that allows evidential ranking of alternative models without direct use of a likelihood function. This approach is particularly important when ranking of several sophisticated stochastic models is desired, and the likelihood is either too complex or impossible to define. We suggest a modification of a Sequential Monte-Carlo sampler that uses ideas of Path Sampling to estimate an approximation to marginal likelihoods as a measure of evidence support. We demonstrate applications of this method on a problem of ranking alternative models of cancerous tumour growth using unique data from three cancerous spheroid lines.
LanguageEnglish
Publication statusPublished - 2012
EventMASAMB - Berlin, Germany
Duration: 10 Apr 201211 Apr 2012

Workshop

WorkshopMASAMB
CountryGermany
CityBerlin
Period10/04/1211/04/12

Fingerprint

Ranking
Alternative models
Marginal likelihood
Tumor
Stochastic model
Sampling
Approximation

Keywords

  • Approximate Bayesian Computation
  • probability
  • Path Sampling

Cite this

Vyshemirsky, V. (2012). Evidential model ranking without likelihoods. MASAMB, Berlin, Germany.
Vyshemirsky, Vladislav. / Evidential model ranking without likelihoods. MASAMB, Berlin, Germany.
@conference{00fc651204e8487eac2e183df48f2679,
title = "Evidential model ranking without likelihoods",
abstract = "We present a probabilistic formulation of the Approximate Bayesian Computation scheme that allows evidential ranking of alternative models without direct use of a likelihood function. This approach is particularly important when ranking of several sophisticated stochastic models is desired, and the likelihood is either too complex or impossible to define. We suggest a modification of a Sequential Monte-Carlo sampler that uses ideas of Path Sampling to estimate an approximation to marginal likelihoods as a measure of evidence support. We demonstrate applications of this method on a problem of ranking alternative models of cancerous tumour growth using unique data from three cancerous spheroid lines.",
keywords = "Approximate Bayesian Computation , probability, Path Sampling",
author = "Vladislav Vyshemirsky",
year = "2012",
language = "English",
note = "MASAMB ; Conference date: 10-04-2012 Through 11-04-2012",

}

Vyshemirsky, V 2012, 'Evidential model ranking without likelihoods' MASAMB, Berlin, Germany, 10/04/12 - 11/04/12, .

Evidential model ranking without likelihoods. / Vyshemirsky, Vladislav.

2012. MASAMB, Berlin, Germany.

Research output: Contribution to conferenceSpeech

TY - CONF

T1 - Evidential model ranking without likelihoods

AU - Vyshemirsky, Vladislav

PY - 2012

Y1 - 2012

N2 - We present a probabilistic formulation of the Approximate Bayesian Computation scheme that allows evidential ranking of alternative models without direct use of a likelihood function. This approach is particularly important when ranking of several sophisticated stochastic models is desired, and the likelihood is either too complex or impossible to define. We suggest a modification of a Sequential Monte-Carlo sampler that uses ideas of Path Sampling to estimate an approximation to marginal likelihoods as a measure of evidence support. We demonstrate applications of this method on a problem of ranking alternative models of cancerous tumour growth using unique data from three cancerous spheroid lines.

AB - We present a probabilistic formulation of the Approximate Bayesian Computation scheme that allows evidential ranking of alternative models without direct use of a likelihood function. This approach is particularly important when ranking of several sophisticated stochastic models is desired, and the likelihood is either too complex or impossible to define. We suggest a modification of a Sequential Monte-Carlo sampler that uses ideas of Path Sampling to estimate an approximation to marginal likelihoods as a measure of evidence support. We demonstrate applications of this method on a problem of ranking alternative models of cancerous tumour growth using unique data from three cancerous spheroid lines.

KW - Approximate Bayesian Computation

KW - probability

KW - Path Sampling

UR - http://masamb2012.molgen.mpg.de/submission_4.html

M3 - Speech

ER -

Vyshemirsky V. Evidential model ranking without likelihoods. 2012. MASAMB, Berlin, Germany.