Author Topic: Propagation of inconsistent evidence has been attempted  (Read 10987 times)

Offline Frans Glassow Tingleff

  • Newbie
  • *
  • Posts: 1
    • View Profile
Propagation of inconsistent evidence has been attempted
« on: March 10, 2014, 10:30:49 »
Hello,

I am currently developing a project in Hugin. I get following error when applying EM-learning in my network:
 
"Propagation of inconsistent evidence has been attempted, or underlow has occured"
 
What does this error-message mean? And what causes it? There is nothing wrong with my data for EM-learning - Could it be that the error is caused by too few observations?

It might be that other users has experienced the same problems.

Offline Nicolaj

  • HUGIN Expert
  • Newbie
  • *****
  • Posts: 5
    • View Profile
Re: Propagation of inconsistent evidence has been attempted
« Reply #1 on: March 11, 2014, 12:10:43 »
Hi Frans

It is a good question. Without having access to your model nor your data I can only give you a qualified guess.

What could have caused the error in your model is that somehow the learning algorithm encounters cases that are impossible according to the probabilities in the model.

Did you manually change the probabilities in the model prior to running the EM learner?
If so, that may be what causes your problems. For instance, in your model you may have A --> B and some state in A which has prior probability 0 or a state in B which has conditional probability 0 given some state in A.

I have been playing around with the EM-Learner and that was the only way I could generate the error message.

Please let us know if this could be a likely explanation.

have a good day!
Nicolaj