def evidence_updated(self, join_tree):
"""
Evidence is updated.
:param join_tree: Join tree.
"""
Propagator.propagate(join_tree)
def evidence_retracted(self, join_tree):
"""
Evidence is retracted.
:param join_tree: Join tree.
"""
Initializer.initialize(join_tree)
Propagator.propagate(join_tree)
def test_propagator():
bbn = BbnUtil.get_huang_graph()
PotentialInitializer.init(bbn)
ug = Moralizer.moralize(bbn)
cliques = Triangulator.triangulate(ug)
join_tree = Transformer.transform(cliques)
Initializer.initialize(join_tree)
Propagator.propagate(join_tree)
def apply_from_serde(join_tree):
"""
Applies propagation to join tree from a deserialzed join tree.
:param join_tree: Join tree.
:return: Join tree (the same one passed in).
"""
join_tree.listener = None
join_tree.evidences = dict()
PotentialInitializer.reinit(join_tree)
Initializer.initialize(join_tree)
Propagator.propagate(join_tree)
join_tree.set_listener(InferenceController())
return join_tree
def apply(bbn):
"""
Sets up the specified BBN for probability propagation in tree clusters (PPTC).
:param bbn: BBN graph.
:return: Join tree.
"""
PotentialInitializer.init(bbn)
ug = Moralizer.moralize(bbn)
cliques = Triangulator.triangulate(ug)
join_tree = Transformer.transform(cliques)
Initializer.initialize(join_tree)
Propagator.propagate(join_tree)
join_tree.set_listener(InferenceController())
return join_tree
def reapply(join_tree, cpts):
"""
Reapply propagation to join tree with new CPTs. The join tree structure is kept but the BBN node CPTs
are updated. A new instance/copy of the join tree will be returned.
:param join_tree: Join tree.
:param cpts: Dictionary of new CPTs. Keys are id's of nodes and values are new CPTs.
:return: Join tree.
"""
jt = copy.deepcopy(join_tree)
jt.update_bbn_cpts(cpts)
jt.listener = None
jt.evidences = dict()
PotentialInitializer.reinit(jt)
Initializer.initialize(jt)
Propagator.propagate(jt)
jt.set_listener(InferenceController())
return jt
def test_propagator():
"""
Tests propagation.
:return: None.
"""
bbn = BbnUtil.get_huang_graph()
PotentialInitializer.init(bbn)
ug = Moralizer.moralize(bbn)
cliques = Triangulator.triangulate(ug)
join_tree = Transformer.transform(cliques)
Initializer.initialize(join_tree)
Propagator.propagate(join_tree)
e_potentials = {
'3=on,4=on,5=on': 0.00315,
'3=on,4=on,5=off': 0.31155,
'3=on,4=off,5=on': 0.00365,
'3=on,4=off,5=off': 0.36165,
'3=off,4=on,5=on': 0.00150,
'3=off,4=on,5=off': 0.14880,
'3=off,4=off,5=on': 0.16800,
'3=off,4=off,5=off': 0.00170,
'4=on,6=on,7=on': 0.00705,
'4=on,6=on,7=off': 0.13395,
'4=on,6=off,7=on': 0.30780,
'4=on,6=off,7=off': 0.01620,
'4=off,6=on,7=on': 0.26030,
'4=off,6=on,7=off': 0.01370,
'4=off,6=off,7=on': 0.24795,
'4=off,6=off,7=off': 0.01305,
'2=on,4=on,6=on': 0.10800,
'2=on,4=on,6=off': 0.02700,
'2=on,4=off,6=on': 0.25200,
'2=on,4=off,6=off': 0.06300,
'2=off,4=on,6=on': 0.03300,
'2=off,4=on,6=off': 0.29700,
'2=off,4=off,6=on': 0.02200,
'2=off,4=off,6=off': 0.19800,
'0=on,1=on,2=on': 0.17500,
'0=off,1=on,2=on': 0.04000,
'0=on,1=on,2=off': 0.07500,
'0=off,1=on,2=off': 0.16000,
'0=on,1=off,2=on': 0.17500,
'0=off,1=off,2=on': 0.06000,
'0=on,1=off,2=off': 0.07500,
'0=off,1=off,2=off': 0.24000,
'1=on,2=on,3=on': 0.19350,
'1=off,2=on,3=on': 0.11750,
'1=on,2=on,3=off': 0.02150,
'1=off,2=on,3=off': 0.11750,
'1=on,2=off,3=on': 0.21150,
'1=off,2=off,3=on': 0.15750,
'1=on,2=off,3=off': 0.02350,
'1=off,2=off,3=off': 0.15750,
'2=on,3=on,4=on': 0.09330,
'2=off,3=on,4=on': 0.22140,
'2=on,3=on,4=off': 0.21770,
'2=off,3=on,4=off': 0.14760,
'2=on,3=off,4=on': 0.04170,
'2=off,3=off,4=on': 0.10860,
'2=on,3=off,4=off': 0.09730,
'2=off,3=off,4=off': 0.07240,
'1=on,2=on': 0.21500,
'1=on,2=off': 0.23500,
'1=off,2=on': 0.23500,
'1=off,2=off': 0.31500,
'2=on,3=on': 0.31100,
'2=on,3=off': 0.13900,
'2=off,3=on': 0.36900,
'2=off,3=off': 0.18100,
'2=on,4=on': 0.13500,
'2=on,4=off': 0.31500,
'2=off,4=on': 0.33000,
'2=off,4=off': 0.22000,
'3=on,4=on': 0.31470,
'3=on,4=off': 0.36530,
'3=off,4=on': 0.15030,
'3=off,4=off': 0.16970,
'4=on,6=on': 0.14100,
'4=on,6=off': 0.32400,
'4=off,6=on': 0.27400,
'4=off,6=off': 0.26100
}
o_potentials = '\n'.join([str(v) for _, v in join_tree.potentials.items()
]).split('\n')
o_potentials = [p.split('|') for p in o_potentials]
o_potentials = {tokens[0]: float(tokens[1]) for tokens in o_potentials}
assert len(e_potentials) == len(o_potentials)
for k, lhs in o_potentials.items():
assert k in e_potentials
rhs = e_potentials[k]
assert lhs == rhs