MN/MRF - To Bayesian Network

Finding a Bayesian Network 𝓑 that is a minimal I-Map for a given Markov Network 𝓜 is:

• conceptually and computationally more difficult than Bayesian Network To Markov Network
• resulting Bayesian Network can be considerably larger

However, if a Markov Network 𝓜 is a chordal graph then there exist a Bayesian Network 𝓑 such that: 𝓜 and 𝓑 are I-Equivalent

Converting a Bayesian Network to Markov Network

the process is not the same (i.e. no one-to-one correspondence)

see: BN - To Markov Network