From: Psy Kosh (email@example.com)
Date: Thu Mar 17 2005 - 13:56:48 MST
First, just wanted to say thanks in general to everyone who had
suggestions for where I ought look.
Next, with regards to the immediate issue...
> IMO, explicit application of Bayes Theorem can play a role in intelligence
> but it certainly can't be the ONLY tool used by an intelligence to figure
> out how to achieve its goals (because of aforementioned computational
> feasibility problems).
As far as I can tell, more or less everyone here agrees on that. Bayes
theorem being sort of an ideal that you should have your inference
algorithims have a sufficiently small error with respect to. (Well, of
course, the ideal is simply getting the right answer...) So a simple
thing may be to somehow estimate what factors may be relavent, what
factors are likely not, then simply do the Bayes thing based on that?
I don't know the best way to do that, of course (which is, in effect,
part of my original question) but near as I can tell there is here a
general acceptance of when doing bayes stuff, it will be "up to the
boundeds of ones computational resouces", of course.
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:50 MDT