ed.get_blanket

Aliases:

• ed.get_blanket
• ed.util.get_blanket

get_blanket(
    x,
    collection=None
)

Defined in edward/util/random_variables.py.

Get Markov blanket of input, which consists of its parents, its children, and the other parents of its children.

Args:

• x: RandomVariable or tf.Tensor. Query node to find Markov blanket of.
• collection: list of RandomVariable. The collection of random variables to check with respect to; defaults to all random variables in the graph.

Returns:

list of RandomVariable. Markov blanket of x.

Examples

a = Normal(0.0, 1.0)
b = Normal(0.0, 1.0)
c = Normal(a * b, 1.0)
d = Normal(0.0, 1.0)
e = Normal(c * d, 1.0)
assert set(ed.get_blanket(c)) == set([a, b, d, e])