0
$\begingroup$

Please helm I am wondering if we consider a learning the parameter of a bayesian network ,with a training set ,where each training set is a vector of values containing all the random variable ,in the network .,Now can we show that the learning parameter of the network according to the MLE ,can be derived by seperately learning the parameters of each conditional distribution

$\endgroup$
  • $\begingroup$ I'm assuming the conditions are completely disjoint and the frequency of each condition is known in advance? $\endgroup$ – barrycarter Nov 8 '14 at 16:46
  • $\begingroup$ yes probability of the marginal distributions are known $\endgroup$ – Susmita Ghosh Nov 13 '14 at 10:00
0
$\begingroup$

Yes, this is explained in the paper Decomposing Parameter Estimation Problems by Refaat et al (2014).

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.