There's a generalization of noisy-or to non-binary variables
and arbitrary deterministic functions (i.e. noisy-f instead of
noisy-or):
Srinivas, Sampath."A generalization of the noisy-or model". UAI-93.
San Franciso: Morgan Kaufmann.
Hope this helps!
Charles
On Tuesday, February 4, 2003, at 05:56 PM, Sean Li wrote:
> Dear all,
>
> Could anybody give me some pointers to the work of the Noise or
> assumption
> applied to non-binary nodes in estimating conditional probabilities?
> Thank
> you very much.
>
> Sean Li
>
--
Charles Sutton
University of Massachusetts, Amherst
[EMAIL PROTECTED]
URL: http://www.cs.umass.edu/~casutton/
