35. Priors as Mathematical Objects

35. Priors as Mathematical Objects

Yudkowsky meint, dass man “Priors”, also alle Informationen die man schon vor einer Beobachtung hat, als mathematische Objekte darstellen kann.

Erscheint mir plausibel aber nicht allzu wichtig.

Naja, wichtiger ist die Frage: was sind Priors genau?

Priors = Prior probability (e.g. how probable is hypothesis H1, in advance of any observations?=P(H1) ) + likelihood ratios (i.e. the relevant conditional probabilities; e.g. how likely are you to see this evidence given the hypothesis H1 is true?= P(E/H1) )

posteriors probability = given the observed evidence, how probable is H1 ? = P(H1/E)



Leave a Reply