Download presentation

Presentation is loading. Please wait.

Published byVanessa Maynard Modified over 2 years ago

1
Bayesian inference Jean Daunizeau Wellcome Trust Centre for Neuroimaging 16 / 05 / 2008

2
Overview of the talk 1 Probabilistic modelling and representation of uncertainty 1.1 Introduction 1.2 Bayesian paradigm 1.3 Specification of priors 1.4 Hierarchical models 2 Numerical Bayesian inference methods 2.1 Sampling methods 2.2 Variational methods (EM, VB) 3 SPM applications 3.1 aMRI segmentation 3.2 fMRI time series analysis with spatial priors 3.3 Dynamic causal modelling 3.4 EEG source reconstruction

3
Overview of the talk 1 Probabilistic modelling and representation of uncertainty 1.1 Introduction 1.2 Bayesian paradigm 1.3 Specification of priors 1.4 Hierarchical models 2 Numerical Bayesian inference methods 2.1 Sampling methods 2.2 Variational methods (EM, VB) 3 SPM applications 3.1 aMRI segmentation 3.2 fMRI time series analysis with spatial priors 3.3 Dynamic causal modelling 3.4 EEG source reconstruction

4
Introduction

5
Overview of the talk 1 Probabilistic modelling and representation of uncertainty 1.1 Introduction 1.2 Bayesian paradigm 1.3 Specification of priors 1.4 Hierarchical models 2 Numerical Bayesian inference methods 2.1 Sampling methods 2.2 Variational methods (EM, VB) 3 SPM applications 3.1 aMRI segmentation 3.2 fMRI time series analysis with spatial priors 3.3 Dynamic causal modelling 3.4 EEG source reconstruction

6
Bayesian paradigm (1) : theory of probability Degree of plausibility desiderata: - should be represented using real numbers(D1) - should conform with intuition(D2) - should be consistent(D3) a=2 b=5 a=2 normalization: marginalization: conditioning : (Bayes rule)

7
Bayesian paradigm (2) : Likelihood and priors generative model m Likelihood: Prior: Bayes rule:

8
Bayesian paradigm (3) : Model comparison Occams razor : Principle of parsimony : « plurality should not be assumed without necessity » model evidence p(y|m) space of all data sets y=f(x) x Model evidence:

9
Overview of the talk 1 Probabilistic modelling and representation of uncertainty 1.1 Introduction 1.2 Bayesian paradigm 1.3 Specification of priors 1.4 Hierarchical models 2 Numerical Bayesian inference methods 2.1 Sampling methods 2.2 Variational methods (EM, VB) 3 SPM applications 3.1 aMRI segmentation 3.2 fMRI time series analysis with spatial priors 3.3 Dynamic causal modelling 3.4 EEG source reconstruction

10
Specification of priors lack of information/entropy order/structure priors = population behaviour / information available before having observed the data subjectivist approach : informative priors objectivist approach : non-informative priors Principle of maximum entropy : find the probability distribution function which maximizes the entropy under some constraints (normalization, expectation, … )

11
Overview of the talk 1 Probabilistic modelling and representation of uncertainty 1.1 Introduction 1.2 Bayesian paradigm 1.3 Specification of priors 1.4 Hierarchical models 2 Numerical Bayesian inference methods 2.1 Sampling methods 2.2 Variational methods (EM, VB) 3 SPM applications 3.1 aMRI segmentation 3.2 fMRI time series analysis with spatial priors 3.3 Dynamic causal modelling 3.4 EEG source reconstruction

12
Hierarchical models (1) : principle hierarchy causality

13
Hierarchical models (2) : directed acyclic graphs (DAGs)

14
Hierarchical models (3) : univariate linear hierarchical model prior posterior

15
Overview of the talk 1 Probabilistic modelling and representation of uncertainty 1.1 Introduction 1.2 Bayesian paradigm 1.3 Specification of priors 1.4 Hierarchical models 2 Numerical Bayesian inference methods 2.1 Sampling methods 2.2 Variational methods (EM, VB) 3 SPM applications 3.1 aMRI segmentation 3.2 fMRI time series analysis with spatial priors 3.3 Dynamic causal modelling 3.4 EEG source reconstruction

16
Overview of the talk 1 Probabilistic modelling and representation of uncertainty 1.1 Introduction 1.2 Bayesian paradigm 1.3 Specification of priors 1.4 Hierarchical models 2 Numerical Bayesian inference methods 2.1 Sampling methods 2.2 Variational methods (EM, VB) 3 SPM applications 3.1 aMRI segmentation 3.2 fMRI time series analysis with spatial priors 3.3 Dynamic causal modelling 3.4 EEG source reconstruction

17
Sampling methods MCMC example: Gibbs sampling

18
Overview of the talk 1 Probabilistic modelling and representation of uncertainty 1.1 Introduction 1.2 Bayesian paradigm 1.3 Specification of priors 1.4 Hierarchical models 2 Numerical Bayesian inference methods 2.1 Sampling methods 2.2 Variational methods (EM, VB) 3 SPM applications 3.1 aMRI segmentation 3.2 fMRI time series analysis with spatial priors 3.3 Dynamic causal modelling 3.4 EEG source reconstruction

19
Variational methods VB/EM/ReML: find (iteratively) the variational posterior q(θ) which maximizes the free energy F(q) under some mean-field approximation:

20
Overview of the talk 1 Probabilistic modelling and representation of uncertainty 1.1 Introduction 1.2 Bayesian paradigm 1.3 Specification of priors 1.4 Hierarchical models 2 Numerical Bayesian inference methods 2.1 Sampling methods 2.2 Variational methods (EM, VB) 3 SPM applications 3.1 aMRI segmentation 3.2 fMRI time series analysis with spatial priors 3.3 Dynamic causal modelling 3.4 EEG source reconstruction

21
Overview of the talk 1 Probabilistic modelling and representation of uncertainty 1.1 Introduction 1.2 Bayesian paradigm 1.3 Specification of priors 1.4 Hierarchical models 2 Numerical Bayesian inference methods 2.1 Sampling methods 2.2 Variational methods (EM, VB) 3 SPM applications 3.1 aMRI segmentation 3.2 fMRI time series analysis with spatial priors 3.3 Dynamic causal modelling 3.4 EEG source reconstruction

22
aMRI segmentation grey matterPPM of belonging to … CSFwhite matter

23
Overview of the talk 1 Probabilistic modelling and representation of uncertainty 1.1 Introduction 1.2 Bayesian paradigm 1.3 Specification of priors 1.4 Hierarchical models 2 Numerical Bayesian inference methods 2.1 Sampling methods 2.2 Variational methods (EM, VB) 3 SPM applications 3.1 aMRI segmentation 3.2 fMRI time series analysis with spatial priors 3.3 Dynamic causal modelling 3.4 EEG source reconstruction

24
fMRI time series analysis with spatial priors observations GLM coeff prior variance of GLM coeff prior variance of data noise AR coeff (correlated noise) ML estimate of W VB estimate of W aMRI smoothed W (RFT)

25
Overview of the talk 1 Probabilistic modelling and representation of uncertainty 1.1 Introduction 1.2 Bayesian paradigm 1.3 Specification of priors 1.4 Hierarchical models 2 Numerical Bayesian inference methods 2.1 Sampling methods 2.2 Variational methods (EM, VB) 3 SPM applications 3.1 aMRI segmentation 3.2 fMRI time series analysis with spatial priors 3.3 Dynamic causal modelling 3.4 EEG source reconstruction

26
Dynamic causal modelling state-space formulation:

27
Overview of the talk 1 Probabilistic modelling and representation of uncertainty 1.1 Introduction 1.2 Bayesian paradigm 1.3 Specification of priors 1.4 Hierarchical models 2 Numerical Bayesian inference methods 2.1 Sampling methods 2.2 Variational methods (EM, VB) 3 SPM applications 3.1 aMRI segmentation 3.2 fMRI time series analysis with spatial priors 3.3 Dynamic causal modelling 3.4 EEG source reconstruction

28
EEG source reconstruction

29
Homo apriorius Homo pragmaticus Homo frequentistus Homo sapiens Homo bayesianis

Similar presentations

© 2016 SlidePlayer.com Inc.

All rights reserved.

Ads by Google