probability adapting

The front end.
Post Reply
nina
Posts: 2
Joined: Tue Mar 21, 2017 3:32 pm

probability adapting

Post by nina »

Hi everyone,
I'm working on a static BN. Now I'd like to adapt the posterior probability of an event xi given that another event Q has been occurred n times, i.e. P(xi|Q = n).

How can I do it using GeNIE? I tried to create the same BN in a dynamic model, then I created an arc of order 1 in the node xi (the one I want to adapt), but actually I don't know how to set the probability update in xi given the evidence and I don't know if this is the right way for the adapting.

Thank you so much
marek [BayesFusion]
Site Admin
Posts: 430
Joined: Tue Dec 11, 2007 4:24 pm

Re: probability adapting

Post by marek [BayesFusion] »

As far a static network goes, I think the way to do it is to create a node Q with several outcomes, e.g., Q0, Q1, Q2_5, Q6_10, each of which denotes the number of times Q has occurred. Then, you enter evidence for the appropriate state. If Q occurred 4 times, for example, the observed state of Q is Q2_5. As far as the influence of the number of times Q has been observed on the event xi goes, you can express this in the conditional probabilities P(Q|xi). I assume, of course, that Q is a child of xi.

You can do this using a DBN. In that case, xi would have Qi as a child. xi-1 would be additionally the parent of xi and would transfer some of the probability that results from prior observations to xi. The magnitude of the influences will depend on the numerical values of conditional probabilities.

Assuming that the occurrence of Q at different time steps is independent of xi, you could also do with with a static BN by creating Q1, Q2, Q3, ..., that are children of xi and then observe them as they come. Multiple Qi observed would increase the probability of xi.
I hope this helps.
Cheers,

Marek
Post Reply