Jeffreys Prior

# Jeffreys prior

to get instant updates about 'Jeffreys Prior' on your MyPage. Meet other similar minded people. Its Free!

X

Description:
In Bayesian probability, the Jeffreys prior, named after Harold Jeffreys, is a non-informative (objective) prior distribution on parameter space that is proportional to the square root of the determinant of the Fisher information:

[itex]pleft(vecthetaright) propto sqrt.,[/itex]

It has the key feature that it is invariant under reparameterization of the parameter vector [itex]vectheta[/itex]. This makes it of special interest for use with scale parametersJaynes, E. T. (1968) "Prior Probabilities", IEEE Trans. on Systems Science and Cybernetics, SSC-4, 227 ..

## Reparameterization

For an alternate parameterization [itex]vecvarphi[/itex] we can derive

[itex]p(vecvarphi) propto sqrt,[/itex]

from

[itex]p(vectheta) propto sqrt,[/itex]

using the change of variables theorem, the definition of Fisher information, and that the product of determinants is the determinant of the matrix product:

[itex]
beginp(vecvarphi) & = p(vectheta) left|detfracright| \& propto sqrt \& = sqrt \& = sqrt \& = sqrt= sqrt.end[/itex]

In the simpler case of a single parameter space variable we can derive

[itex]
beginp(varphi) & = p(theta) left|fracright|propto sqrt \& = sqrt= sqrt= sqrt.end[/itex]

## Attributes

From a practical and mathematical standpoint, a valid reason to use this non-informative prior...

No feeds found

All