def get_matches(self, x, constraints=None, epsilon=0.0):
if self.operator is None:
return
# print("GETTING MATCHES")
grounded = [(ground(a), x[a]) for a in x if (isinstance(a, tuple))]
# print("FACTS")
# pprint(grounded)
index = build_index(grounded)
# print("INDEX")
# pprint(index)
# print("OPERATOR")
# pprint(self.operator)
for m in self.operator.match(index, epsilon=epsilon):
# print('match', m, self.operator.name)
result = tuple(
[unground(subst(m, ele)) for ele in self.operator.name[1:]])
# result = tuple(['?' + subst(m, ele)
# for ele in self.operator.name[1:]])
# result = tuple(['?' + m[e] for e in self.target_types])
# print('GET MATCHES T', result)
yield result
def get_matches(self, x, epsilon=0.0):
# print("GETTING MATCHES")
# pprint(self.learner.get_hset())
grounded = self.ground_example(x)
# grounded = [(ground(a), x[a]) for a in x if (isinstance(a, tuple))]
# print("FACTS")
# pprint(grounded)
index = build_index(grounded)
# print("INDEX")
# pprint(index)
# print("OPERATOR")
# pprint(self.operator)
# print(self.learner.get_hset())
for h in self.learner.get_hset():
# print('h', h)
# Update to include initial args
operator = Operator(tuple(('Rule', ) + self.args), h, [])
# print("OPERATOR", h)
for m in operator.match(index, epsilon=epsilon):
# print('match', m, operator.name)
result = tuple(
[unground(subst(m, ele)) for ele in operator.name[1:]])
# result = tuple(['?' + subst(m, ele)
# for ele in self.operator.name[1:]])
# result = tuple(['?' + m[e] for e in self.target_types])
# print('GET MATCHES T', result)
yield result
def check_match(self, t, x):
"""
if y is predicted to be 1 then returns True
else returns False
"""
# print("CHECK MATCHES T", t)
if self.operator is None:
return
t = tuple(ele.replace('?', "QM") for ele in t)
mapping = {a: t[i] for i, a in enumerate(self.operator.name[1:])}
# print("MY MAPPING", mapping)
# print("CHECKING MATCHES")
grounded = [(ground(a), x[a]) for a in x if (isinstance(a, tuple))]
# pprint(grounded)
# pprint(mapping)
index = build_index(grounded)
for m in self.operator.match(index, initial_mapping=mapping):
return True
return False
def get_matches(self, x, epsilon=0.0):
# print("GETTING MATCHES")
# pprint(self.tuples)
grounded = self.ground_example(x)
# grounded = [(ground(a), x[a]) for a in x if (isinstance(a, tuple))]
# print("FACTS")
# pprint(grounded)
index = build_index(grounded)
# print("INDEX")
# pprint(index)
# print("OPERATOR")
# pprint(self.operator)
# print(self.learner.get_hset())
for t in self.tuples:
mapping = {a: t[i] for i, a in enumerate(self.args)}
operator = Operator(tuple(('Rule', ) + self.args),
frozenset().union(self.constraints), [])
for m in operator.match(index,
epsilon=epsilon,
initial_mapping=mapping):
result = tuple(ele.replace("QM", '?') for ele in t)
# print('GET MATCHES T', result)
yield result
break
def check_match(self, t, x):
# print("CHECK MATCHES T", t)
t = tuple(ground(ele) for ele in t)
# Update to include initial args
mapping = {a: t[i] for i, a in enumerate(self.args)}
# print("MY MAPPING", mapping)
# print("CHECKING MATCHES")
if t not in self.tuples:
return False
grounded = self.ground_example(x)
# grounded = [(ground(a), x[a]) for a in x if (isinstance(a, tuple))]
# pprint(grounded)
# pprint(mapping)
index = build_index(grounded)
# Update to include initial args
operator = Operator(tuple(('Rule', ) + self.args),
frozenset().union(self.constraints), [])
for m in operator.match(index, initial_mapping=mapping):
return True
return False
def covers(h, x, initial_mapping):
"""
Returns true if h covers x
"""
index = build_index(x)
operator = Operator(tuple(['Rule']), h, [])
for m in operator.match(index, initial_mapping=initial_mapping):
return True
return False
def get_matches(self, x, epsilon=0.0):
x = x.get_view("flat_ungrounded")
grounded = self.ground_example(x)
index = build_index(grounded)
for t in self.tuples:
mapping = {a: t[i] for i, a in enumerate(self.args)}
operator = Operator(tuple(('Rule', ) + self.args),
frozenset().union(self.constraints), [])
for m in operator.match(index,
epsilon=epsilon,
initial_mapping=mapping):
result = tuple(ele.replace("QM", '?') for ele in t)
yield result
break
def is_specialization(self, s, h):
"""
Takes two hypotheses s and g and returns True if s is a specialization
of h. Note, it returns False if s and h are equal (s is not a
specialization in this case).
"""
if s == h:
return False
# remove vars, so the unification isn't going in both directions.
s = set(remove_vars(l) for l in s)
# check if h matches s (then s specializes h)
index = build_index(s)
operator = Operator(tuple(['Rule']), h, [])
for m in operator.match(index):
return True
return False
def check_match(self, t, x):
x = x.get_view("flat_ungrounded")
# print("CHECK MATCHES T", t)
t = tuple(ground(ele) for ele in t)
# Update to include initial args
mapping = {a: t[i] for i, a in enumerate(self.args)}
if t not in self.tuples:
return False
grounded = self.ground_example(x)
index = build_index(grounded)
# Update to include initial args
operator = Operator(tuple(('Rule', ) + self.args),
frozenset().union(self.constraints), [])
for m in operator.match(index, initial_mapping=mapping):
return True
return False
def check_match(self, t, x):
# print("CHECK MATCHES T", t)
t = tuple(ele.replace('?', "QM") for ele in t)
# Update to include initial args
mapping = {a: t[i] for i, a in enumerate(self.args)}
# print("MY MAPPING", mapping)
# print("CHECKING MATCHES")
grounded = self.ground_example(x)
# grounded = [(ground(a), x[a]) for a in x if (isinstance(a, tuple))]
# pprint(grounded)
# pprint(mapping)
index = build_index(grounded)
for h in self.learner.get_hset():
# Update to include initial args
operator = Operator(tuple(('Rule', ) + self.args), h, [])
for m in operator.match(index, initial_mapping=mapping):
return True
return False
def optimize_clause(h, constraints, seed, pset, nset):
"""
Returns the set of most specific generalization of h that do NOT
cover x.
"""
c_length = clause_length(h)
print('# POS', len(pset))
print('# NEG', len(nset))
print('length', c_length)
p_covered, n_covered = test_coverage(h, constraints, pset, nset)
p_uncovered = [p for p in pset if p not in p_covered]
n_uncovered = [n for n in nset if n not in n_covered]
print("P COVERED", len(p_covered))
print("N COVERED", len(n_covered))
initial_score = clause_score(clause_accuracy_weight, len(p_covered),
len(p_uncovered), len(n_covered),
len(n_uncovered), c_length)
p, pm = seed
pos_partial = list(compute_bottom_clause(p, pm))
# print('POS PARTIAL', pos_partial)
# TODO if we wanted we could add the introduction of new variables to the
# get_variablizations function.
possible_literals = {}
for i, l in enumerate(pos_partial):
possible_literals[i] = [None, l] + [v for v in get_variablizations(l)]
partial_literals = set(
[l for i in possible_literals for l in possible_literals[i]])
additional_literals = h - partial_literals
if len(additional_literals) > 0:
p_index = build_index(p)
operator = Operator(tuple(('Rule', )), h.union(constraints), [])
for add_m in operator.match(p_index, initial_mapping=pm):
break
additional_lit_mapping = {
rename(add_m, l): l
for l in additional_literals
}
for l in additional_lit_mapping:
new_l = additional_lit_mapping[l]
reverse_m = {pm[a]: a for a in pm}
l = rename(reverse_m, l)
print(pos_partial)
print(add_m)
print(l)
print(new_l)
print(additional_literals)
if l not in pos_partial:
print("ERROR l not in pos_partial")
import time
time.sleep(1000)
possible_literals[pos_partial.index(l)].append(new_l)
# pprint(possible_literals)
reverse_pl = {
l: (i, j)
for i in possible_literals for j, l in enumerate(possible_literals[i])
}
clause_vector = [0 for i in range(len(possible_literals))]
for l in h:
if l not in reverse_pl:
print("MISSING LITERAL!!!")
import time
time.sleep(1000)
i, j = reverse_pl[l]
clause_vector[i] = j
clause_vector = tuple(clause_vector)
# print("INITIAL CLAUSE VECTOR")
# print(clause_vector)
flip_weights = [(len(possible_literals[i]) - 1, i)
for i in possible_literals]
size = 1
for w, _ in flip_weights:
size *= (w + 1)
print("SIZE OF SEARCH SPACE:", size)
num_successors = sum([w for w, c in flip_weights])
print("NUM SUCCESSORS", num_successors)
temp_length = num_successors
temp_length = 10
# initial_temp = 0.15
# initial_temp = 0.0
print("TEMP LENGTH", temp_length)
print('INITIAL SCORE', initial_score)
problem = ClauseOptimizationProblem(clause_vector,
initial_cost=-1 * initial_score,
extra=(possible_literals, flip_weights,
constraints, pset, nset,
len(p_uncovered),
len(n_covered)))
# for sol in hill_climbing(problem):
for sol in simulated_annealing(
problem,
# initial_temp=initial_temp,
temp_length=temp_length):
# print("SOLUTION FOUND", sol.state)
print('FINAL SCORE', -1 * sol.cost())
return build_clause(sol.state, possible_literals)
def successors(self, node):
h = node.state
# print("EXPANDING H", h)
args, constraints, pset, neg, neg_mapping, gensym = node.extra
all_args = set(s for x in h.union(constraints)
for s in extract_strings(x) if is_variable(s))
if len(pset) == 0:
return
p, pm = choice(pset)
p_index = build_index(p)
operator = Operator(tuple(('Rule', ) + tuple(all_args)),
h.union(constraints), [])
# operator = Operator(tuple(('Rule',) + args), h, [])
found = False
for m in operator.match(p_index, initial_mapping=pm):
reverse_m = {m[a]: a for a in m}
pos_partial = set([rename(reverse_m, x) for x in p])
found = True
break
if not found:
return
n_index = build_index(neg)
found = False
for nm in operator.match(n_index, initial_mapping=neg_mapping):
# print(nm)
reverse_nm = {nm[a]: a for a in nm}
neg_partial = set([rename(reverse_nm, x) for x in neg])
found = True
break
if not found:
return
unique_pos = pos_partial - neg_partial
unique_neg = neg_partial - pos_partial
# print("UNIQUE POS", unique_pos)
# print("UNIQUE NEG", unique_neg)
# Yield all minimum specializations of current vars
for a in m:
# TODO make sure m[a] is a minimum specialization
sub_m = {a: m[a]}
new_h = frozenset([subst(sub_m, ele) for ele in h])
# print("SPECIALIZATION", new_h, sub_m)
# print()
yield Node(new_h, node, ('specializing', (a, m[a])),
node.cost() + 1, node.extra)
# Add Negations for all neg specializations
# for a in nm:
# sub_nm = {a: nm[a]}
# new_nh = set()
# for ele in h:
# new = subst(sub_nm, ele)
# if new != ele and new not in h:
# new_nh.add(('not', new))
# new_h = h.union(new_nh)
# print("NEGATION SPECIALIZATION", new_nh)
# yield Node(new_h, node, ('negation specialization', (a, nm[a])),
# node.cost()+1, node.extra)
# if current vars then add all relations that include current vars
if len(all_args) > 0:
added = set()
for literal in unique_pos:
if literal in h or literal in constraints:
continue
args = set(s for s in extract_strings(literal)
if is_variable(s))
if len(args.intersection(all_args)) > 0:
key = (literal[0], ) + tuple(
ele if is_variable(ele) else '?'
for ele in literal[1:])
if key in added:
continue
added.add(key)
literal = generalize_literal(literal, gensym)
new_h = h.union(frozenset([literal]))
# print("ADD CURRENT", new_h)
# print()
yield Node(new_h, node, ('adding current', literal),
node.cost() + 1, node.extra)
else:
added = set()
for literal in unique_pos:
if literal in h or literal in constraints:
continue
if literal[0] in added:
continue
added.add(literal[0])
literal = generalize_literal(literal, gensym)
new_h = h.union(frozenset([literal]))
# print("ADD NEW", new_h)
# print()
yield Node(new_h, node, ('adding', literal),
node.cost() + 1, node.extra)
