Jethro's Braindump

Non-informative Priors

Non-informative priors are used in Bayesian Inference, in the situation where no prior information exists, or inference based dominantly on the data is desired. We wish to find a distribution p(θ) that contained “no information” about θ (favours no value of θ over another).

An example of such a non-informative prior in the discrete space would be a uniform distribution over θ.

Links to this note