def test_keys(self):
# A regression test to make sure issue #49 does no longer pop up. By
# iterating over the configuration in the for loop, it should not raise
# a KeyError if the child hyperparameter is inactive.
cs = ConfigurationSpace()
shrinkage = CategoricalHyperparameter(
"shrinkage",
["None", "auto", "manual"],
default_value="None",
)
shrinkage_factor = UniformFloatHyperparameter(
"shrinkage_factor",
0.,
1.,
0.5,
)
cs.add_hyperparameters([shrinkage, shrinkage_factor])
cs.add_condition(EqualsCondition(shrinkage_factor, shrinkage,
"manual"))
for i in range(10):
config = cs.sample_configuration()
{
hp_name: config[hp_name]
for hp_name in config if config[hp_name] is not None
}
def test_acts_as_mapping(self):
"""
Test that ConfigurationSpace can act as a mapping with iteration,
indexing and items, values, keys.
"""
cs = ConfigurationSpace()
names = [f"name{i}" for i in range(5)]
hyperparameters = [
UniformIntegerHyperparameter(name, 0, 10) for name in names
]
cs.add_hyperparameters(hyperparameters)
# Test indexing
assert cs['name3'] == hyperparameters[3]
# Test dict methods
assert list(cs.keys()) == names
assert list(cs.values()) == hyperparameters
assert list(cs.items()) == list(zip(names, hyperparameters))
assert len(cs) == 5
# Test __iter__
assert list(iter(cs)) == names
# Test unpacking
d = {**cs}
assert list(d.keys()) == names
assert list(d.values()) == hyperparameters
assert list(d.items()) == list(zip(names, hyperparameters))
assert len(d) == 5
def test_check_neighbouring_config_diamond_str(self):
diamond = ConfigurationSpace()
head = CategoricalHyperparameter('head', ['red', 'green'])
left = CategoricalHyperparameter('left', ['red', 'green'])
right = CategoricalHyperparameter('right', ['red', 'green', 'blue', 'yellow'])
bottom = CategoricalHyperparameter('bottom', ['red', 'green'])
diamond.add_hyperparameters([head, left, right, bottom])
diamond.add_condition(EqualsCondition(left, head, 'red'))
diamond.add_condition(EqualsCondition(right, head, 'red'))
diamond.add_condition(AndConjunction(EqualsCondition(bottom, left, 'green'),
EqualsCondition(bottom, right, 'green')))
config = Configuration(diamond, {'bottom': 'red', 'head': 'red', 'left': 'green', 'right': 'green'})
hp_name = "head"
index = diamond.get_idx_by_hyperparameter_name(hp_name)
neighbor_value = 1
new_array = ConfigSpace.c_util.change_hp_value(
diamond,
config.get_array(),
hp_name,
neighbor_value,
index
)
expected_array = np.array([1, np.nan, np.nan, np.nan])
np.testing.assert_almost_equal(new_array, expected_array)
def get_hyperparameter_search_space(dataset_properties=None):
cs = ConfigurationSpace()
BaseImageClassificationModel.set_training_space(cs)
BaseImageClassificationModel.set_optimizer_space(cs)
res_kernel_size = CategoricalHyperparameter('res_kernel_size', [3, 5],
default_value=3)
res_stage2_block = UniformIntegerHyperparameter('res_stage2_block',
1,
3,
default_value=2)
res_stage3_block = UniformIntegerHyperparameter('res_stage3_block',
1,
11,
default_value=3)
res_stage4_block = UniformIntegerHyperparameter('res_stage4_block',
1,
47,
default_value=5)
res_stage5_block = UniformIntegerHyperparameter('res_stage5_block',
1,
3,
default_value=2)
cs.add_hyperparameters([
res_kernel_size, res_stage2_block, res_stage3_block,
res_stage4_block, res_stage5_block
])
return cs
def get_pipeline_config_space(self, algorithm_candidates):
cs = ConfigurationSpace()
estimator_choice = CategoricalHyperparameter(
"estimator",
algorithm_candidates,
default_value=algorithm_candidates[0])
cs.add_hyperparameter(estimator_choice)
if self.task_type == IMG_CLS:
aug_space = get_aug_hyperparameter_space()
cs.add_hyperparameters(aug_space.get_hyperparameters())
cs.add_conditions(aug_space.get_conditions())
for estimator_id in algorithm_candidates:
sub_cs = self.get_model_config_space(estimator_id,
include_estimator=False,
include_aug=False)
parent_hyperparameter = {
'parent': estimator_choice,
'value': estimator_id
}
cs.add_configuration_space(
estimator_id,
sub_cs,
parent_hyperparameter=parent_hyperparameter)
return cs
def test_sobol(self):
cs = ConfigurationSpace()
hyperparameters = [
UniformFloatHyperparameter('x%d' % (i + 1), 0, 1)
for i in range(21201)
]
cs.add_hyperparameters(hyperparameters)
sobol_kwargs = dict(
rng=np.random.RandomState(1),
traj_logger=unittest.mock.Mock(),
ta_run_limit=1000,
configs=None,
n_configs_x_params=None,
max_config_fracs=0.25,
init_budget=1,
)
SobolDesign(cs=cs, **sobol_kwargs).select_configurations()
cs.add_hyperparameter(UniformFloatHyperparameter('x21202', 0, 1))
with self.assertRaisesRegex(
Exception,
"Maximum supported dimensionality is 21201.",
):
SobolDesign(cs=cs, **sobol_kwargs).select_configurations()
def _get_configuration_space(self) -> ConfigurationSpace:
"""Get the configuration space for the random forest.
Returns
-------
ConfigurationSpace
"""
cfg = ConfigurationSpace()
cfg.seed(int(self.rs.randint(0, 1000)))
num_trees = Constant("num_trees", value=N_TREES)
bootstrap = CategoricalHyperparameter(
"do_bootstrapping",
choices=(self.bootstrap, ),
default_value=self.bootstrap,
)
max_feats = CategoricalHyperparameter("max_features",
choices=(3 / 6, 4 / 6, 5 / 6, 1),
default_value=1)
min_split = UniformIntegerHyperparameter("min_samples_to_split",
lower=1,
upper=10,
default_value=2)
min_leavs = UniformIntegerHyperparameter("min_samples_in_leaf",
lower=1,
upper=10,
default_value=1)
cfg.add_hyperparameters(
[num_trees, bootstrap, max_feats, min_split, min_leavs])
return cfg
def get_hyperparameter_search_space():
cs = ConfigurationSpace()
n_estimators = UniformIntegerHyperparameter(name="n_estimators",
lower=50,
upper=500,
default_value=50,
log=False)
learning_rate = UniformFloatHyperparameter(name="learning_rate",
lower=0.01,
upper=2,
default_value=0.1,
log=True)
algorithm = CategoricalHyperparameter(name="algorithm",
choices=["SAMME.R", "SAMME"],
default_value="SAMME.R")
max_depth = UniformIntegerHyperparameter(name="max_depth",
lower=2,
upper=8,
default_value=3,
log=False)
cs.add_hyperparameters(
[n_estimators, learning_rate, algorithm, max_depth])
return cs
def test_check_neighbouring_config_diamond_str(self):
diamond = ConfigurationSpace()
head = CategoricalHyperparameter('head', ['red', 'green'])
left = CategoricalHyperparameter('left', ['red', 'green'])
right = CategoricalHyperparameter('right',
['red', 'green', 'blue', 'yellow'])
bottom = CategoricalHyperparameter('bottom', ['red', 'green'])
diamond.add_hyperparameters([head, left, right, bottom])
diamond.add_condition(EqualsCondition(left, head, 'red'))
diamond.add_condition(EqualsCondition(right, head, 'red'))
diamond.add_condition(
AndConjunction(EqualsCondition(bottom, left, 'green'),
EqualsCondition(bottom, right, 'green')))
config = Configuration(diamond, {
'bottom': 'red',
'head': 'red',
'left': 'green',
'right': 'green'
})
hp_name = "head"
index = diamond.get_idx_by_hyperparameter_name(hp_name)
neighbor_value = 1
new_array = ConfigSpace.c_util.change_hp_value(diamond,
config.get_array(),
hp_name, neighbor_value,
index)
expected_array = np.array([1, np.nan, np.nan, np.nan])
np.testing.assert_almost_equal(new_array, expected_array)
def test_check_neighbouring_config_diamond(self):
diamond = ConfigurationSpace()
head = CategoricalHyperparameter('head', [0, 1])
left = CategoricalHyperparameter('left', [0, 1])
right = CategoricalHyperparameter('right', [0, 1, 2, 3])
bottom = CategoricalHyperparameter('bottom', [0, 1])
diamond.add_hyperparameters([head, left, right, bottom])
diamond.add_condition(EqualsCondition(left, head, 0))
diamond.add_condition(EqualsCondition(right, head, 0))
diamond.add_condition(
AndConjunction(EqualsCondition(bottom, left, 1),
EqualsCondition(bottom, right, 1)))
config = Configuration(diamond, {
'bottom': 0,
'head': 0,
'left': 1,
'right': 1
})
hp_name = "head"
index = diamond.get_idx_by_hyperparameter_name(hp_name)
neighbor_value = 1
new_array = change_hp_value(diamond, config.get_array(), hp_name,
neighbor_value, index)
expected_array = np.array([1, np.nan, np.nan, np.nan])
np.testing.assert_almost_equal(new_array, expected_array)
def get_configs(self, variables):
from ConfigSpace import ConfigurationSpace
from ConfigSpace.hyperparameters import UniformFloatHyperparameter
from ConfigSpace.hyperparameters import CategoricalHyperparameter
api_config = {}
cs = ConfigurationSpace()
hp_list = []
for var in variables:
if isinstance(var, ContinuousVariable):
api_config[var.name] = {
'type': 'real',
'space': 'linear',
'range': var.domain
}
hp_list.append(
UniformFloatHyperparameter(name=var.name,
lower=var.domain[0],
upper=var.domain[1],
log=False))
else:
api_config[var.name] = {'type': 'cat', 'values': var.domain}
hp_list.append(
CategoricalHyperparameter(name=var.name,
choices=var.domain))
cs.add_hyperparameters(hp_list)
return api_config, cs
def test_fix_types(self):
# Test categorical and ordinal
for hyperparameter_type in [CategoricalHyperparameter, OrdinalHyperparameter]:
cs = ConfigurationSpace()
cs.add_hyperparameters([
hyperparameter_type('bools', [True, False]),
hyperparameter_type('ints', [1, 2, 3, 4, 5]),
hyperparameter_type('floats', [1.5, 2.5, 3.5, 4.5, 5.5]),
hyperparameter_type('str', ['string', 'ding', 'dong']),
hyperparameter_type('mixed', [2, True, 1.5, 'string', False, 'False']),
])
c = cs.get_default_configuration().get_dictionary()
# Check bools
for b in [False, True]:
c['bools'] = b
c_str = {k: str(v) for k, v in c.items()}
self.assertEqual(fix_types(c_str, cs), c)
# Check legal mixed values
for m in [2, True, 1.5, 'string']:
c['mixed'] = m
c_str = {k: str(v) for k, v in c.items()}
self.assertEqual(fix_types(c_str, cs), c)
# Check error on cornercase that cannot be caught
for m in [False, 'False']:
c['mixed'] = m
c_str = {k: str(v) for k, v in c.items()}
self.assertRaises(ValueError, fix_types, c_str, cs)
# Test constant
for m in [2, 1.5, 'string']:
cs = ConfigurationSpace()
cs.add_hyperparameter(Constant('constant', m))
c = cs.get_default_configuration().get_dictionary()
c_str = {k: str(v) for k, v in c.items()}
self.assertEqual(fix_types(c_str, cs), c)
def get_hyperparameter_search_space():
"""
['n_estimators', 'learning_rate', 'max_depth', 'colsample_bytree', 'gamma',
'min_child_weight', 'reg_alpha', 'reg_lambda', 'subsample']
"""
cs = ConfigurationSpace()
n_estimators = UniformIntegerHyperparameter("n_estimators",
100,
1000,
q=10,
default_value=500)
learning_rate = UniformFloatHyperparameter("learning_rate",
1e-3,
0.9,
log=True,
default_value=0.1)
max_depth = UniformIntegerHyperparameter("max_depth", 1, 12)
colsample_bytree = UniformFloatHyperparameter("colsample_bytree",
0.1,
1,
q=0.1,
default_value=1)
gamma = UniformFloatHyperparameter("gamma",
0,
10,
q=0.1,
default_value=0)
min_child_weight = UniformFloatHyperparameter("min_child_weight",
0,
10,
q=0.1,
default_value=1)
reg_alpha = UniformFloatHyperparameter("reg_alpha",
0,
10,
q=0.1,
default_value=0)
reg_lambda = UniformFloatHyperparameter("reg_lambda",
1,
10,
q=0.1,
default_value=1)
subsample = UniformFloatHyperparameter("subsample",
0.1,
1,
q=0.1,
default_value=1)
cs.add_hyperparameters([
n_estimators, max_depth, learning_rate, min_child_weight,
subsample, colsample_bytree, gamma, reg_alpha, reg_lambda
])
return cs
def get_configspace(self, optimizer='smac'):
if optimizer == 'smac':
cs = ConfigurationSpace()
penalty = CategoricalHyperparameter(
"penalty", ["l1", "l2"], default_value="l2")
loss = CategoricalHyperparameter(
"loss", ["hinge", "squared_hinge"], default_value="squared_hinge")
dual = CategoricalHyperparameter("dual", ['True', 'False'], default_value='True')
# This is set ad-hoc
tol = UniformFloatHyperparameter(
"tol", 1e-5, 1e-1, default_value=1e-4, log=True)
C = UniformFloatHyperparameter(
"C", 0.03125, 32768, log=True, default_value=1.0)
multi_class = Constant("multi_class", "ovr")
# These are set ad-hoc
fit_intercept = Constant("fit_intercept", "True")
intercept_scaling = Constant("intercept_scaling", 1)
cs.add_hyperparameters([penalty, loss, dual, tol, C, multi_class,
fit_intercept, intercept_scaling])
penalty_and_loss = ForbiddenAndConjunction(
ForbiddenEqualsClause(penalty, "l1"),
ForbiddenEqualsClause(loss, "hinge")
)
constant_penalty_and_loss = ForbiddenAndConjunction(
ForbiddenEqualsClause(dual, "False"),
ForbiddenEqualsClause(penalty, "l2"),
ForbiddenEqualsClause(loss, "hinge")
)
penalty_and_dual = ForbiddenAndConjunction(
ForbiddenEqualsClause(dual, "True"),
ForbiddenEqualsClause(penalty, "l1")
)
cs.add_forbidden_clause(penalty_and_loss)
cs.add_forbidden_clause(constant_penalty_and_loss)
cs.add_forbidden_clause(penalty_and_dual)
return cs
elif optimizer == 'tpe':
from hyperopt import hp
space = {'penalty': hp.choice('liblinear_combination',
[{'penalty': "l1", 'loss': "squared_hinge", 'dual': "False"},
{'penalty': "l2", 'loss': "hinge", 'dual': "True"},
{'penalty': "l2", 'loss': "squared_hinge", 'dual': "True"},
{'penalty': "l2", 'loss': "squared_hinge", 'dual': "False"}]),
'loss': None,
'dual': None,
'tol': hp.loguniform('liblinear_tol', np.log(1e-5), np.log(1e-1)),
'C': hp.loguniform('liblinear_C', np.log(0.03125), np.log(32768)),
'multi_class': hp.choice('liblinear_multi_class', ["ovr"]),
'fit_intercept': hp.choice('liblinear_fit_intercept', ["True"]),
'intercept_scaling': hp.choice('liblinear_intercept_scaling', [1])}
return space
else:
raise ValueError('Unknown optimizer %s when getting configspace' % optimizer)
def get_configspace(self, optimizer='smac'):
if optimizer == 'smac':
cs = ConfigurationSpace()
cs.add_hyperparameters([
UniformFloatHyperparameter("x%s" % i, self.lb, self.ub)
for i in range(1, self.dim + 1)
])
return cs
else:
raise ValueError('Unknown optimizer %s when getting configspace' %
optimizer)
def get_configuration_space(self):
cs = ConfigurationSpace()
finite_horizon = CategoricalHyperparameter(name="finite_horizon",
choices=["true", "false"], default_value="true")
horizon = UniformIntegerHyperparameter(name="horizon",
lower=1, upper=1000, default_value=10)
use_horizon = CSC.InCondition(child=horizon, parent=finite_horizon,
values=["true"])
cs.add_hyperparameters([horizon, finite_horizon])
cs.add_condition(use_horizon)
return cs
def set_training_space(cs: ConfigurationSpace):
'''
Set hyperparameters for training
'''
batch_size = CategoricalHyperparameter('batch_size', [16, 32],
default_value=32)
keep_prob = UniformFloatHyperparameter('keep_prob',
0,
0.99,
default_value=0.5)
cs.add_hyperparameters([batch_size, keep_prob])
def main(argv):
if len(argv) < 2:
raise ValueError("You should input filename.")
if FLAGS.processes < 1 or FLAGS.processes > cpu_count():
raise ValueError("Please input a proper number of processes.")
# 读取输入文件
if argv[1].endswith(".csv"):
# 如果后缀名为csv则按csv读取
df = pd.read_csv(sys.argv[1])
X = df[df.columns[:-1]].values
y = df[df.columns[-1]].values
else:
# 否则按照libsvm读取
with open(argv[1], "r") as fp:
lines = fp.readlines()
X, y = libsvm2sparse(lines)
# 分割数据集
X_train, X_valid, y_train, y_valid = train_test_split(X,
y,
test_size=0.33,
random_state=1)
processes = FLAGS.processes
# 指定输入输出目录
dirs = [
"tmpfile/smac3-output_%s" % (datetime.datetime.fromtimestamp(
time.time()).strftime('%Y-%m-%d_%H:%M:%S_%f'))
for _ in range(processes)
]
# 创建进程池
pool = []
for i in range(FLAGS.processes):
cs = ConfigurationSpace()
# 超参搜索空间，使用[1e-6, 1]
alpha = UniformFloatHyperparameter(name="alpha",
lower=1e-3,
upper=1,
default_value=(i + 1) / processes,
log=False)
cs.add_hyperparameters([alpha])
# 指定并行pSMAC目录和当前的输出目录
pool.append(
ParallelSMBO(X_train,
y_train,
X_valid,
y_valid,
dirs=dirs,
smbo_id=i,
cs=cs,
our_work=FLAGS.our_work))
for i in range(processes):
pool[i].start()
for i in range(processes):
pool[i].join()
def set_optimizer_space(cs: ConfigurationSpace):
'''
Set hyperparameters for optimizers
'''
optimizer = CategoricalHyperparameter('optimizer', ['SGD', 'Adam'],
default_value='Adam')
sgd_lr = UniformFloatHyperparameter('sgd_lr',
0.00001,
0.1,
default_value=0.005,
log=True) # log scale
sgd_decay = UniformFloatHyperparameter('sgd_decay',
0.0001,
0.1,
default_value=0.05,
log=True) # log scale
sgd_momentum = UniformFloatHyperparameter('sgd_momentum',
0.3,
0.99,
default_value=0.9)
adam_lr = UniformFloatHyperparameter('adam_lr',
0.00001,
0.1,
default_value=0.005,
log=True) # log scale
adam_decay = UniformFloatHyperparameter('adam_decay',
0.0001,
0.1,
default_value=0.05,
log=True) # log scale
sgd_lr_cond = InCondition(child=sgd_lr,
parent=optimizer,
values=['SGD'])
sgd_decay_cond = InCondition(child=sgd_decay,
parent=optimizer,
values=['SGD'])
sgd_momentum_cond = InCondition(child=sgd_momentum,
parent=optimizer,
values=['SGD'])
adam_lr_cond = InCondition(child=adam_lr,
parent=optimizer,
values=['Adam'])
adam_decay_cond = InCondition(child=adam_decay,
parent=optimizer,
values=['Adam'])
cs.add_hyperparameters(
[optimizer, sgd_lr, sgd_decay, sgd_momentum, adam_lr, adam_decay])
cs.add_conditions([
sgd_lr_cond, sgd_decay_cond, sgd_momentum_cond, adam_lr_cond,
adam_decay_cond
])
def get_hyperparameter_search_space(dataset_properties=None):
cs = ConfigurationSpace()
BaseImageClassificationModel.set_training_space(cs)
BaseImageClassificationModel.set_optimizer_space(cs)
vgg_kernel_size = CategoricalHyperparameter('vgg_kernel_size', [3, 5], default_value=3)
vgg_keep_prob = UniformFloatHyperparameter('vgg_keep_prob', 0, 0.99, default_value=0.5)
vgg_block2_layer = UniformIntegerHyperparameter('vgg_block2_layer', 2, 3, default_value=2)
vgg_block3_layer = UniformIntegerHyperparameter('vgg_block3_layer', 2, 5, default_value=3)
vgg_block4_layer = UniformIntegerHyperparameter('vgg_block4_layer', 2, 5, default_value=3)
vgg_block5_layer = UniformIntegerHyperparameter('vgg_block5_layer', 2, 5, default_value=3)
cs.add_hyperparameters(
[vgg_kernel_size, vgg_keep_prob, vgg_block2_layer, vgg_block3_layer, vgg_block4_layer, vgg_block5_layer])
return cs
def get_hyperparameter_search_space():
cs = ConfigurationSpace()
BaseImageClassificationModel.set_training_space(cs)
BaseImageClassificationModel.set_optimizer_space(cs)
inceptionv3_block_a = UniformIntegerHyperparameter(
'inceptionv3_block_a', 2, 4, default_value=3)
inceptionv3_block_b = UniformIntegerHyperparameter(
'inceptionv3_block_b', 3, 5, default_value=4)
inceptionv3_block_c = UniformIntegerHyperparameter(
'inceptionv3_block_c', 1, 3, default_value=2)
cs.add_hyperparameters(
[inceptionv3_block_a, inceptionv3_block_b, inceptionv3_block_c])
return cs
def get_configspace(self, optimizer='smac'):
if optimizer == 'smac':
cs = ConfigurationSpace()
cs.add_hyperparameters(
[UniformFloatHyperparameter("x%s" % i, self.lb, self.ub) for i in range(1, 1 + 2)])
return cs
elif optimizer == 'gpflowopt':
import gpflowopt
domain = gpflowopt.domain.ContinuousParameter('x1', self.lb, self.ub) + \
gpflowopt.domain.ContinuousParameter('x2', self.lb, self.ub)
return domain
else:
raise ValueError('Unknown optimizer %s when getting configspace' % optimizer)
def get_configspace(self, optimizer='smac'):
if optimizer == 'smac':
cs = ConfigurationSpace()
x1 = UniformFloatHyperparameter("x1", -2.25, 2.5)
x2 = UniformFloatHyperparameter("x2", -2.5, 1.75)
cs.add_hyperparameters([x1, x2])
return cs
elif optimizer == 'gpflowopt':
import gpflowopt
domain = gpflowopt.domain.ContinuousParameter('x1', -2.25, 2.5) + \
gpflowopt.domain.ContinuousParameter('x2', -2.5, 1.75)
return domain
else:
raise ValueError('Unknown optimizer %s when getting configspace' % optimizer)
def test_remove_inactive_parameter():
configuration_space = ConfigurationSpace(seed=1)
hp1 = CategoricalHyperparameter("hp1", choices=[0, 1])
hp2 = CategoricalHyperparameter("hp2", choices=['a'])
hp3 = UniformIntegerHyperparameter("hp3",
lower=0,
upper=5,
default_value=5)
configuration_space.add_hyperparameters([hp1, hp2, hp3])
# If hp1 = 0, then don't allow hp2
not_condition = NotEqualsCondition(hp2, hp1, 0)
configuration_space.add_condition(not_condition)
allowed_cfg = Configuration(configuration_space, {
'hp1': 1,
'hp2': 'a',
'hp3': 5
})
not_allowed = {'hp1': 0, 'hp2': 'a', 'hp3': 5}
with pytest.raises(ValueError):
Configuration(configuration_space, not_allowed)
# No inactive hp - case: config is CS.configuration
transformed = AbstractBenchmark._check_and_cast_configuration(
allowed_cfg, configuration_space)
assert transformed.get_dictionary() == {'hp1': 1, 'hp2': 'a', 'hp3': 5}
# No inactive hp - case: config is dict
transformed = AbstractBenchmark._check_and_cast_configuration(
allowed_cfg.get_dictionary(), configuration_space)
assert transformed.get_dictionary() == {'hp1': 1, 'hp2': 'a', 'hp3': 5}
# Remove inactive: - case: config is CS.configuration
not_allowed_cs = Configuration(configuration_space, {
'hp1': 0,
'hp2': 'a',
'hp3': 5
},
allow_inactive_with_values=True)
transformed = AbstractBenchmark._check_and_cast_configuration(
not_allowed_cs, configuration_space)
assert transformed.get_dictionary() == {'hp1': 0, 'hp3': 5}
# Remove inactive: - case: config is dict
transformed = AbstractBenchmark._check_and_cast_configuration(
not_allowed, configuration_space)
assert transformed.get_dictionary() == {'hp1': 0, 'hp3': 5}
def get_hyperparameter_search_space(dataset_properties=None):
# TODO add hyperparameter to gbdt binning
cs = ConfigurationSpace()
binning_method = CategoricalHyperparameter()
# shrinkage = UniformFloatHyperparameter(
# name="shrinkage", lower=0.0, upper=1.0, default_value=0.5
# )
# n_components = UniformIntegerHyperparameter(
# name="n_components", lower=1, upper=29, default_value=10
# )
# tol = UniformFloatHyperparameter(
# name="tol", lower=0.0001, upper=1, default_value=0.0001
# )
cs.add_hyperparameters([binning_method])
return cs
class SVMSpace(ParamSpace):
def __init__(self):
super().__init__()
self.name = "SVM"
self.model = LinearSVC
self.is_deterministic = False
self.configuration_space = ConfigurationSpace()
self.configuration_space.add_hyperparameters([
UniformFloatHyperparameter("tol", 1e-5, 1e-1, default_value=1e-4, log=True),
UniformFloatHyperparameter("C", 0.03125, 32768, log=True, default_value=1.0)
])
def _initialize_algorithm(self, random_state=None, **config):
return self.model(penalty="l2", loss="squared_hinge", dual=False, multi_class="ovr", fit_intercept=True,
intercept_scaling=1, **config)
def get_hyperparameter_search_space():
cs = ConfigurationSpace()
n_estimators = UniformIntegerHyperparameter("n_estimators",
100,
1000,
q=50,
default_value=500)
num_leaves = UniformIntegerHyperparameter("num_leaves",
31,
2047,
default_value=128)
max_depth = UnParametrizedHyperparameter('max_depth', 15)
learning_rate = UniformFloatHyperparameter("learning_rate",
1e-3,
0.9,
log=True,
default_value=0.1)
min_child_weight = UniformFloatHyperparameter("min_child_weight",
0,
10,
q=0.1,
default_value=1)
subsample = UniformFloatHyperparameter("subsample",
0.1,
1,
q=0.1,
default_value=1)
colsample_bytree = UniformFloatHyperparameter("colsample_bytree",
0.1,
1,
q=0.1,
default_value=1)
reg_alpha = UniformFloatHyperparameter("reg_alpha",
0,
10,
q=0.1,
default_value=0)
reg_lambda = UniformFloatHyperparameter("reg_lambda",
1,
10,
q=0.1,
default_value=1)
cs.add_hyperparameters([
n_estimators, num_leaves, max_depth, learning_rate,
min_child_weight, subsample, colsample_bytree, reg_alpha,
reg_lambda
])
return cs
def test_keys(self):
# A regression test to make sure issue #49 does no longer pop up. By
# iterating over the configuration in the for loop, it should not raise
# a KeyError if the child hyperparameter is inactive.
cs = ConfigurationSpace()
shrinkage = CategoricalHyperparameter(
"shrinkage", ["None", "auto", "manual"], default_value="None",
)
shrinkage_factor = UniformFloatHyperparameter(
"shrinkage_factor", 0., 1., 0.5,
)
cs.add_hyperparameters([shrinkage, shrinkage_factor])
cs.add_condition(EqualsCondition(shrinkage_factor, shrinkage, "manual"))
for i in range(10):
config = cs.sample_configuration()
{hp_name: config[hp_name] for hp_name in config if config[hp_name] is not None}
class DecisionTreeSpace(ParamSpace):
def __init__(self):
super().__init__()
self.name = "DecisionTree"
self.model = DecisionTreeClassifier
self.is_deterministic = False
self.configuration_space = ConfigurationSpace()
self.configuration_space.add_hyperparameters([
CategoricalHyperparameter("criterion", ["gini", "entropy"], default_value="gini"),
UniformIntegerHyperparameter('max_depth', 1, 20, default_value=20),
UniformIntegerHyperparameter("min_samples_split", 2, 20, default_value=2),
UniformIntegerHyperparameter("min_samples_leaf", 1, 20, default_value=1)
])
def _initialize_algorithm(self, random_state=None, **config):
return self.model(min_weight_fraction_leaf=0, max_features=1.0, max_leaf_nodes=None,
min_impurity_decrease=0.0, random_state=random_state,
**config)
class GradientBoostingSpace(ParamSpace):
def __init__(self):
super().__init__()
self.name = "GBM"
self.model = LGBMClassifier
self.is_deterministic = True
self.configuration_space = ConfigurationSpace()
self.configuration_space.add_hyperparameters([
UniformIntegerHyperparameter("num_leaves", 4, 64, default_value=32),
UniformIntegerHyperparameter("min_child_samples", 1, 100, default_value=20),
UniformIntegerHyperparameter("max_depth", 3, 12, default_value=12),
UniformFloatHyperparameter("reg_alpha", 0, 1, default_value=0),
UniformFloatHyperparameter("reg_lambda", 0, 1, default_value=0),
CategoricalHyperparameter('boosting_type', choices=["gbdt", "dart", "goss"])
])
def _initialize_algorithm(self, random_state=None, **config):
return self.model(n_estimators=100, verbose=-1, n_jobs=-1, random_state=random_state, **config)
class RandomForestSpace(ParamSpace):
def __init__(self):
super().__init__()
self.name = "RandomForest"
self.model = LGBMClassifier
self.is_deterministic = False
self.configuration_space = ConfigurationSpace()
self.configuration_space.add_hyperparameters([
UniformFloatHyperparameter("colsample_bytree", 0.20, 0.80, default_value=0.70),
UniformFloatHyperparameter("subsample", 0.20, 0.80, default_value=0.66),
UniformIntegerHyperparameter("num_leaves", 4, 64, default_value=32),
UniformIntegerHyperparameter("min_child_samples", 1, 100, default_value=20),
UniformIntegerHyperparameter("max_depth", 4, 12, default_value=12),
])
def _initialize_algorithm(self, random_state=None, **config):
return self.model(n_estimators=100, subsample_freq=1, boosting_type="rf", verbose=-1, n_jobs=-1,
random_state=random_state, **config)
def test_deactivate_inactive_hyperparameters(self):
diamond = ConfigurationSpace()
head = CategoricalHyperparameter('head', [0, 1])
left = CategoricalHyperparameter('left', [0, 1])
right = CategoricalHyperparameter('right', [0, 1])
bottom = CategoricalHyperparameter('bottom', [0, 1])
diamond.add_hyperparameters([head, left, right, bottom])
diamond.add_condition(EqualsCondition(left, head, 0))
diamond.add_condition(EqualsCondition(right, head, 0))
diamond.add_condition(AndConjunction(EqualsCondition(bottom, left, 0),
EqualsCondition(bottom, right, 0)))
c = deactivate_inactive_hyperparameters({'head': 0, 'left': 0,
'right': 0, 'bottom': 0},
diamond)
diamond._check_configuration_rigorous(c)
c = deactivate_inactive_hyperparameters({'head': 1, 'left': 0,
'right': 0, 'bottom': 0},
diamond)
diamond._check_configuration_rigorous(c)
c = deactivate_inactive_hyperparameters({'head': 0, 'left': 1,
'right': 0, 'bottom': 0},
diamond)
diamond._check_configuration_rigorous(c)
diamond = ConfigurationSpace()
head = CategoricalHyperparameter('head', [0, 1])
left = CategoricalHyperparameter('left', [0, 1])
right = CategoricalHyperparameter('right', [0, 1])
bottom = CategoricalHyperparameter('bottom', [0, 1])
diamond.add_hyperparameters([head, left, right, bottom])
diamond.add_condition(EqualsCondition(left, head, 0))
diamond.add_condition(EqualsCondition(right, head, 0))
diamond.add_condition(OrConjunction(EqualsCondition(bottom, left, 0),
EqualsCondition(bottom, right, 0)))
c = deactivate_inactive_hyperparameters({'head': 0, 'left': 0,
'right': 0, 'bottom': 0},
diamond)
diamond._check_configuration_rigorous(c)
c = deactivate_inactive_hyperparameters({'head': 1, 'left': 1,
'right': 0, 'bottom': 0},
diamond)
diamond._check_configuration_rigorous(c)
c = deactivate_inactive_hyperparameters({'head': 0, 'left': 1,
'right': 0, 'bottom': 0},
diamond)
diamond._check_configuration_rigorous(c)
plain = ConfigurationSpace()
a = UniformIntegerHyperparameter('a', 0, 10)
b = UniformIntegerHyperparameter('b', 0, 10)
plain.add_hyperparameters([a, b])
c = deactivate_inactive_hyperparameters({'a': 5, 'b': 6}, plain)
plain.check_configuration(c)
def test_deactivate_inactive_hyperparameters(self):
diamond = ConfigurationSpace()
head = CategoricalHyperparameter('head', [0, 1])
left = CategoricalHyperparameter('left', [0, 1])
right = CategoricalHyperparameter('right', [0, 1])
bottom = CategoricalHyperparameter('bottom', [0, 1])
diamond.add_hyperparameters([head, left, right, bottom])
diamond.add_condition(EqualsCondition(left, head, 0))
diamond.add_condition(EqualsCondition(right, head, 0))
diamond.add_condition(AndConjunction(EqualsCondition(bottom, left, 0),
EqualsCondition(bottom, right, 0)))
c = deactivate_inactive_hyperparameters({'head': 0, 'left': 0,
'right': 0, 'bottom': 0},
diamond)
diamond._check_configuration_rigorous(c)
c = deactivate_inactive_hyperparameters({'head': 1, 'left': 0,
'right': 0, 'bottom': 0},
diamond)
diamond._check_configuration_rigorous(c)
c = deactivate_inactive_hyperparameters({'head': 0, 'left': 1,
'right': 0, 'bottom': 0},
diamond)
diamond._check_configuration_rigorous(c)
diamond = ConfigurationSpace()
head = CategoricalHyperparameter('head', [0, 1])
left = CategoricalHyperparameter('left', [0, 1])
right = CategoricalHyperparameter('right', [0, 1])
bottom = CategoricalHyperparameter('bottom', [0, 1])
diamond.add_hyperparameters([head, left, right, bottom])
diamond.add_condition(EqualsCondition(left, head, 0))
diamond.add_condition(EqualsCondition(right, head, 0))
diamond.add_condition(OrConjunction(EqualsCondition(bottom, left, 0),
EqualsCondition(bottom, right, 0)))
c = deactivate_inactive_hyperparameters({'head': 0, 'left': 0,
'right': 0, 'bottom': 0},
diamond)
diamond._check_configuration_rigorous(c)
c = deactivate_inactive_hyperparameters({'head': 1, 'left': 1,
'right': 0, 'bottom': 0},
diamond)
diamond._check_configuration_rigorous(c)
c = deactivate_inactive_hyperparameters({'head': 0, 'left': 1,
'right': 0, 'bottom': 0},
diamond)
diamond._check_configuration_rigorous(c)
plain = ConfigurationSpace()
a = UniformIntegerHyperparameter('a', 0, 10)
b = UniformIntegerHyperparameter('b', 0, 10)
plain.add_hyperparameters([a, b])
c = deactivate_inactive_hyperparameters({'a': 5, 'b': 6}, plain)
plain.check_configuration(c)