Hi,
I am new to working with smile and genie.
I have created a simple graphical model in genie with 4 nodes and each node has a defined set of outcomes with some probability values. Now I have imported this model to the .net framework using the smile.dll library. I would like to set evidence along with their probability values dynamically runtime using smile wrappers. Smile provides a wrapper called network.setEvidence() to set evidence. But I am not able to find a function that helps me update the evidence along with the probability values.
Kindly help me in this regard.
Best Regards
Phanindra
update probability values for an independent node created in genie
-
- Posts: 2
- Joined: Tue Jul 03, 2018 2:30 pm
-
- Site Admin
- Posts: 1423
- Joined: Mon Nov 26, 2007 5:51 pm
Re: update probability values for an independent node created in genie
If you change the evidence, you should call Network.UpdateBeliefs to recalculate the probabilities.
-
- Posts: 2
- Joined: Tue Jul 03, 2018 2:30 pm
Re: update probability values for an independent node created in genie
Dear Shooltz,shooltz[BayesFusion] wrote:If you change the evidence, you should call Network.UpdateBeliefs to recalculate the probabilities.
Many thanks for your reply. I would like to update the evidence along with the probabilities. Please consider the below example
let us say I have a testNetwork with node name weather, possible evidence are sunny with 0.5 probability and rainy with 0.5 probability created in genie. now I have imported this testNetowrk into .net using smile read file function. I would like to set the evidence for the node weather with possible evidence as sunny with probability 0.75 and rainy with probability 0.25. then I would like to invoke the Network.UpdateBeliefs function to recalculate the probabilities.
Kindly share your comments on the same.
Best Regards
Phanindra
-
- Site Admin
- Posts: 1423
- Joined: Mon Nov 26, 2007 5:51 pm
Re: update probability values for an independent node created in genie
Try using Network.SetVirtualEvidence. See Using GeNIe/Bayesian networks/Virtual Evidence chapter in GeNIe's manual for the theoretical background on virtual evidence.