Dear UAI List Members,
  I would like to seek your assistance in identifying the latest research
or papers related to training a BN with missing data for the
intermediate nodes. The BN structure is given and the training data is only
related to the top and leave nodes. There are no observations for the
intermediate nodes. The question is how to learn the CPTs for the
intermediate nodes. The conventional EM method does not seem to work
in this case.

Thanks,

QJ

Reply via email to