Fisher Information Approximation

#Bayesian #Model Selection #FIA

FIA is a method to describe the minimum description length ( MDL MDL is a measure of how well a model compresses data by minimizing the combined cost of the description of the model and the misfit. ) of models,

$$ \mathrm{FIA} = -\ln p(y | \hat\theta) + \frac{k}{2} \ln \frac{n}{2\pi} + \ln \int_\Theta \sqrt{ \operatorname{det}[I(\theta)] d\theta } $$

  • $I(\theta)$: Fisher information matrix of sample size 1.
  • $$I_{i,j}(\theta) = E\left( \frac{\partial \ln p(y| \theta)}{\partial \theta_i}\frac{ \partial \ln p (y | \theta) }{ \partial \theta_j } \right)$$.

Published: by ;

Lei Ma (2020). 'Fisher Information Approximation', Datumorphism, 11 April. Available at: https://datumorphism.leima.is/cards/statistics/fia/.

Table of Contents

Current Ref:

  • cards/statistics/fia.md