Conditional Independence (CI) graph is a special type of a Probabilistic
Graphical Model (PGM) where the feature connections are modeled using an
undirected graph and the edge weights show the partial correlation strength
between the features. Since the CI graphs capture direct depende