## Weight and interval parameters

The engine.
Gary
Posts: 40
Joined: Thu Nov 17, 2016 2:38 am

### Weight and interval parameters

How to use smile to generate and evaluate images with weight or interval parameters？ Is there some procedural statement for reference?
thank you!

marek [BayesFusion]
Posts: 299
Joined: Tue Dec 11, 2007 4:24 pm

### Re: Weight and interval parameters

Could you please clarify your question? What do you mean by images? What do you mean by weights? There are several parameters that can be called weights, e.g., parameters in linear models, weight in utility functions, etc. Do you mean specification of interval rather than point probabilities?
Cheers,

Marek

Gary
Posts: 40
Joined: Thu Nov 17, 2016 2:38 am

### Re: Weight and interval parameters

I'm sorry. it's my fault . For example, there are four nodes, two chance nodes, a decision node, a utility node, how to use smile to create a weighted image of the qualitative image? I have built the Influence graph，there are operation of '+'and 'x' , I do not know how to write the statement to use them and generate a qualitative impact map .
thank you!

Gary
Posts: 40
Joined: Thu Nov 17, 2016 2:38 am

### Re: Weight and interval parameters

The '+' rules is 'AddRule[4][4]={{'+'.'-'.'0','?'},{'?','-','-','?'},{'+','-','0','?'},{'?','?','?','?'}} and the 'X' rules is ' MutipleRule[4][4]={{'+'.'-'.'0','?'},{'-'.'+'.'0','?'},{'0'.'0.'0','0'},{'?','?','0','?'}} i don't know how to use it by smile.

marek [BayesFusion]
Posts: 299
Joined: Tue Dec 11, 2007 4:24 pm

### Re: Weight and interval parameters

OK, I'm starting to understand what you want to do. It seems that you want to build a qualitative probabilistic network (like a QPN) with signs of influences being propagated rather than numbers. GeNIe and SMILE are quantitative, i.e., they require a full numerical specification of the joint probability distribution. You can simulate the behavior of qualitative networks but please note that there are infinitely many quantitative networks for every qualitative network, i.e., there are many possible sets of parameters that will fulfill the qualitative constraints. Furthermore, when a QPN shows a ?, a quantitative network will be much more precise and will effectively show an increase or a decrease in the probability (i.e., + or -). Am I interpreting your question correctly?

If so, I'm afraid GeNIe cannot help you much. We used to have QGeNIe when still at the University of Pittsburgh but we have not turned it into a product yet. QGeNIe is still quantitative but it makes use of canonical gates (DeMorgan gate) and allows for specifying each influence by one number and shows the posteriors by colors rather than numbers. Very useful in initial stages of model building or in group decision-making sessions. Here is a paper describing QGeNIe: http://www.pitt.edu/~druzdzel/psfiles/imcsit09.pdf. We will release it as a product within months, so please stay tuned.
I hope this helps.

Marek

Gary
Posts: 40
Joined: Thu Nov 17, 2016 2:38 am

### Re: Weight and interval parameters

Thank you very much!