Markov Random Fields

---

Resources:

• https://www.youtube.com/watch?v=iBQkZdPHlCs
• page=235 Ch. 12.3

---

Any two subsets $S$ and $T$ of variables are conditionally independent given a separating subset

The separating subset in this case this can be $\{B\},\{B,D\},\{B,E\},\{B,D,E\}$ (blocks the path between $S$ and $T$).

• If you observe any of these subsets, this makes $S$ and $T$ conditionally independent.

Bayesian Networks as MRFs

We need to add an edge to the MRF (moralizing the parent). However we lose information doing this (since $A$ and $B$ are supposed to be independent of eachother if C is not given)