Dear Roldano:
>
> Dear collegues
>
> I am looking for references about the use of contraints in building
> Bayesian Belief Networks.
> For example let us suppose to have as prior knowledge (1) a prior network
> as a first aproximation of the target network and (2) a set of constraints
> on the values of the variables in particular cases. Is it possible to
> refine the prior network using the constraints to obtain a new networks
> matching such constraints?
> One way could be to adopt some learning methods, using the constraints to
> create the data base for the learning process; but is there an alternative
> way to refine directly the conditional ditributions according to the
> constraints?
I analyzed some problems of that kind in:
Valtorta, M. and Donald W. Loveland.
"On the Complexity of Belief Network Synthesis and Refinement."
_International Journal of Approximate Reasoning_, 7 (1992), pp.121-148.
>
>
> Many thanks in advance
>
> Yours Sincerely
>
> Roldano Cattoni
>
>
> +----------------------------------------------------------------+
> | |
> | Roldano Cattoni |
> | SSI Division ------ __o __o |
> | I.R.S.T. - Pante' di Povo ----- _`\<,_ _`\<,_ |
> | 38050 Trento - ITALY ----- (*)/ (*) (*)/ (*) |
> | |
> | WWW: http://hera.itc.it:2002/~cattoni/ |
> | e-mail: [EMAIL PROTECTED] |
> | tel: +39+0461+314.547 (or 444) |
> | fax: +39+0461+314.591 |
> | secretariat: +39+0461+314.592 (or 517) |
> | |
> +----------------------------------------------------------------+
>
You are welcome, and good luck!
Marco
Marco Valtorta, Associate Professor and Undergraduate Director
Department of Computer Science internet: [EMAIL PROTECTED]
University of South Carolina tel.: (1)(803)777-4641 fax: -3767
Columbia, SC 29208, U.S.A. http://www.cs.sc.edu/~mgv/ tlx: 805038 USC
---------------------------------------------------------------------------
"Probability is not about numbers. It is about the structure of reasoning."
--Glenn Shafer
---------------------------------------------------------------------------
