def __init__(self, dim=2, bounds=None,
noise_std=0, random_state=None):
self.dim = dim
params = {'x%d' % i: (0, 10, 5) for i in range(1, 1+self.dim)}
config_space = ConfigurationSpace()
config_space.add_hyperparameters([UniformFloatHyperparameter(k, *v) for k, v in params.items()])
super().__init__(config_space, noise_std, optimal_value=0, random_state=random_state)
def get_cs_bc():
cs_bc = ConfigurationSpace()
x0 = UniformFloatHyperparameter("x0", scale1[0], scale1[1])
# x0 = UniformIntegerHyperparameter("x0", scale1[0], scale1[1]) # test int
x1 = UniformFloatHyperparameter("x1", scale2[0], scale2[1])
cs_bc.add_hyperparameters([x0, x1])
return cs_bc
def __init__(self, noise_std=0, random_state=None):
params = {'x%d' % i: (-1.25, 1.25, 1) for i in [1, 2]}
config_space = ConfigurationSpace()
config_space.add_hyperparameters([UniformFloatHyperparameter(k, *v) for k, v in params.items()])
super().__init__(config_space, noise_std,
optimal_value=-0.072,
optimal_point=[(0.84852813, -0.84852813), (-0.84852813, 0.84852813)],
random_state=random_state)
def __init__(self, noise_std=0, random_state=None):
params = {'x1': (-10, 0, -5), 'x2': (-6.5, 0, -3.25)}
config_space = ConfigurationSpace()
config_space.add_hyperparameters([UniformFloatHyperparameter(k, *v) for k, v in params.items()])
super().__init__(config_space, noise_std,
optimal_value=-106.7645367,
optimal_point=[(-3.1302468, -1.5821422)],
random_state=random_state)
def __init__(self, noise_std=0, random_state=None):
lb, ub = -4.5, 4.5
dim = 2
params = {'x%d' % i: (lb, ub, (lb + ub)/2)
for i in range(1, dim+1)}
config_space = ConfigurationSpace()
config_space.add_hyperparameters([UniformFloatHyperparameter(k, *v) for k, v in params.items()])
super().__init__(config_space, noise_std, optimal_value=0, random_state=random_state)
def __init__(self, noise_std=0, random_state=None):
params = {'x1': (-5, 10, 0),
'x2': (0, 15, 0)}
config_space = ConfigurationSpace()
config_space.add_hyperparameters([UniformFloatHyperparameter(k, *v) for k, v in params.items()])
super().__init__(config_space, noise_std,
optimal_value=0.397887,
optimal_point=[(-np.pi, 12.275), (np.pi, 2.275), (9.42478, 2.475)],
random_state=random_state)
def __init__(self, noise_std=0, random_state=None):
self.ref_point = [1864.72022, 11.81993945, 0.2903999384]
params = {'x%d' % i: (1.0, 3.0) for i in range(1, 6)}
config_space = ConfigurationSpace()
config_space.add_hyperparameters([UniformFloatHyperparameter(k, *v) for k, v in params.items()])
super().__init__(config_space, noise_std,
num_objs=3,
random_state=random_state)
def __init__(self, noise_std=0, random_state=None):
params = {'x1': (-15.0, -5.0, -10.0),
'x2': (-3.0, 3.0, 0)}
config_space = ConfigurationSpace()
config_space.add_hyperparameters([UniformFloatHyperparameter(k, *v) for k, v in params.items()])
super().__init__(config_space, noise_std,
optimal_value=0,
optimal_point=[(-10.0, 1.0)],
random_state=random_state)
def __init__(self, dim: int, num_constraints=0, noise_std=0, random_state=None):
self.dim = dim
self.ref_point = [11.0, 11.0]
params = {'x%d' % i: (0, 1) for i in range(1, dim+1)}
config_space = ConfigurationSpace()
config_space.add_hyperparameters([UniformFloatHyperparameter(k, *v) for k, v in params.items()])
super().__init__(config_space, noise_std,
num_objs=2, num_constraints=num_constraints,
random_state=random_state)
def __init__(self, dim=2, constrained=False, noise_std=0, random_state=None):
self.dim = dim
self.constrained = constrained
params = {'x%d' % i: (-5.0, 10.0, 2.5) for i in range(1, 1+self.dim)}
config_space = ConfigurationSpace()
config_space.add_hyperparameters([UniformFloatHyperparameter(k, *v) for k, v in params.items()])
super().__init__(config_space, noise_std,
optimal_value=0,
optimal_point=[tuple(1.0 for _ in range(self.dim))],
random_state=random_state)
def __init__(self, noise_std=0, random_state=None):
self.ref_point = [10.0, 10.0]
params = {'x1': (0.1, 10.0),
'x2': (0.0, 5.0)}
config_space = ConfigurationSpace()
config_space.add_hyperparameters([UniformFloatHyperparameter(k, *v) for k, v in params.items()])
super().__init__(config_space, noise_std,
num_objs=2, num_constraints=2,
random_state=random_state)
def get_xgboost_config_space(task_type='cls'):
if task_type == 'cls':
cs = ConfigurationSpace()
n_estimators = UniformIntegerHyperparameter("n_estimators",
100,
1000,
q=50,
default_value=500)
max_depth = UniformIntegerHyperparameter("max_depth", 1, 12)
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)
gamma = UniformFloatHyperparameter("gamma",
0,
10,
q=0.1,
default_value=0)
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, max_depth, learning_rate, min_child_weight,
subsample, colsample_bytree, gamma, reg_alpha, reg_lambda
])
return cs
elif task_type == 'rgs':
raise NotImplementedError
else:
raise ValueError('Unsupported task type: %s.' % (task_type, ))
def __init__(self, constrained=False, noise_std=0, random_state=None):
self.ref_point = [18.0, 6.0]
self.constrained = constrained
num_constraints = 1 if self.constrained else 0
params = {'x1': (0, 1, 0.5),
'x2': (0, 1, 0.5)}
config_space = ConfigurationSpace()
config_space.add_hyperparameters([UniformFloatHyperparameter(k, *v) for k, v in params.items()])
super().__init__(config_space, noise_std,
num_objs=2,
num_constraints=num_constraints,
random_state=random_state)
def __init__(self, dim, num_objs=2, num_constraints=0, noise_std=0, random_state=None):
if dim <= num_objs:
raise ValueError(
"dim must be > num_objs, but got %s and %s" % (dim, num_objs)
)
self.dim = dim
self.k = self.dim - num_objs + 1
self.bounds = [(0.0, 1.0) for _ in range(self.dim)]
self.ref_point = [self._ref_val for _ in range(num_objs)]
params = {'x%d' % i: (0, 1, i/dim) for i in range(1, dim+1)}
config_space = ConfigurationSpace()
config_space.add_hyperparameters([UniformFloatHyperparameter(k, *v) for k, v in params.items()])
super().__init__(config_space, noise_std, num_objs, num_constraints, random_state=random_state)
def __init__(self, dim=2, bounds=None, constrained=False,
noise_std=0, random_state=None):
self.constrained = constrained
if bounds is None:
if constrained:
lb, ub = -5, 10
else:
lb, ub = -10, 15
else:
lb, ub = bounds
params = {'x%d' % i: (lb, ub, (lb + ub)/2)
for i in range(1, dim+1)}
config_space = ConfigurationSpace()
config_space.add_hyperparameters([UniformFloatHyperparameter(k, *v) for k, v in params.items()])
super().__init__(config_space, noise_std, optimal_value=0, random_state=random_state)
def get_hyperparameter_search_space(space_size='large'):
"""
['n_estimators', 'learning_rate', 'max_depth', 'colsample_bytree', 'gamma',
'min_child_weight', 'reg_alpha', 'reg_lambda', 'subsample']
"""
cs = ConfigurationSpace()
if space_size == 'large':
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)
elif space_size == 'medium':
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 = UnParametrizedHyperparameter("reg_alpha", 0)
reg_lambda = UnParametrizedHyperparameter("reg_lambda", 1)
subsample = UnParametrizedHyperparameter("subsample", 1)
else:
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 = UnParametrizedHyperparameter("colsample_bytree", 1)
gamma = UnParametrizedHyperparameter("gamma", 0)
min_child_weight = UnParametrizedHyperparameter("min_child_weight", 1)
reg_alpha = UnParametrizedHyperparameter("reg_alpha", 0)
reg_lambda = UnParametrizedHyperparameter("reg_lambda", 1)
subsample = UnParametrizedHyperparameter("subsample", 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_lightgbm_config_space(task_type='cls'):
if task_type == 'cls':
cs = ConfigurationSpace()
n_estimators = UniformIntegerHyperparameter("n_estimators",
100,
1000,
default_value=500,
q=50)
num_leaves = UniformIntegerHyperparameter("num_leaves",
31,
2047,
default_value=128)
max_depth = Constant('max_depth', 15)
learning_rate = UniformFloatHyperparameter("learning_rate",
1e-3,
0.3,
default_value=0.1,
log=True)
min_child_samples = UniformIntegerHyperparameter("min_child_samples",
5,
30,
default_value=20)
subsample = UniformFloatHyperparameter("subsample",
0.7,
1,
default_value=1,
q=0.1)
colsample_bytree = UniformFloatHyperparameter("colsample_bytree",
0.7,
1,
default_value=1,
q=0.1)
cs.add_hyperparameters([
n_estimators, num_leaves, max_depth, learning_rate,
min_child_samples, subsample, colsample_bytree
])
return cs
elif task_type == 'rgs':
raise NotImplementedError
else:
raise ValueError('Unsupported task type: %s.' % (task_type, ))
def get_cs_lightgbm(): # todo q and int for compare?
cs = ConfigurationSpace()
n_estimators = UniformFloatHyperparameter("n_estimators",
100,
1000,
default_value=500,
q=50)
num_leaves = UniformIntegerHyperparameter("num_leaves",
31,
2047,
default_value=128)
# max_depth = Constant('max_depth', 15)
learning_rate = UniformFloatHyperparameter("learning_rate",
1e-3,
0.3,
default_value=0.1,
log=True)
min_child_samples = UniformIntegerHyperparameter("min_child_samples",
5,
30,
default_value=20)
subsample = UniformFloatHyperparameter("subsample",
0.7,
1,
default_value=1,
q=0.1)
colsample_bytree = UniformFloatHyperparameter("colsample_bytree",
0.7,
1,
default_value=1,
q=0.1)
# cs.add_hyperparameters([n_estimators, num_leaves, max_depth, learning_rate, min_child_samples, subsample,
# colsample_bytree])
cs.add_hyperparameters([
n_estimators, num_leaves, learning_rate, min_child_samples,
subsample, colsample_bytree
])
return cs
logging.basicConfig(level=logging.INFO)
# Build Configuration Space which defines all parameters and their ranges
cs = ConfigurationSpace()
# We define a few possible types of SVM-kernels and add them as "kernel" to our cs
kernel = CategoricalHyperparameter("kernel",
["linear", "rbf", "poly", "sigmoid"],
default_value="poly")
cs.add_hyperparameter(kernel)
# There are some hyperparameters shared by all kernels
C = UniformFloatHyperparameter("C", 0.001, 1000.0, default_value=1.0)
shrinking = CategoricalHyperparameter("shrinking", ["true", "false"],
default_value="true")
cs.add_hyperparameters([C, shrinking])
# Others are kernel-specific, so we can add conditions to limit the searchspace
degree = UniformIntegerHyperparameter(
"degree", 1, 5, default_value=3) # Only used by kernel poly
coef0 = UniformFloatHyperparameter("coef0", 0.0, 10.0,
default_value=0.0) # poly, sigmoid
cs.add_hyperparameters([degree, coef0])
use_degree = InCondition(child=degree, parent=kernel, values=["poly"])
use_coef0 = InCondition(child=coef0, parent=kernel, values=["poly", "sigmoid"])
cs.add_conditions([use_degree, use_coef0])
# This also works for parameters that are a mix of categorical and values from a range of numbers
# For example, gamma can be either "auto" or a fixed float
gamma = CategoricalHyperparameter(
"gamma", ["auto", "value"],
result = dict()
result['objs'] = [
t1 + t2 + t3,
]
result['constraints'] = [
np.sum((X + 5)**2) - 25,
]
return result
if __name__ == "__main__":
params = {'float': {'x0': (-10, 0, -5), 'x1': (-6.5, 0, -3.25)}}
cs = ConfigurationSpace()
cs.add_hyperparameters([
UniformFloatHyperparameter(name, *para)
for name, para in params['float'].items()
])
bo = SMBO(mishra,
cs,
num_constraints=1,
num_objs=1,
acq_optimizer_type='random_scipy',
max_runs=50,
time_limit_per_trial=10,
task_id='soc')
history = bo.run()
print(history)
history.plot_convergence(true_minimum=-106.7645367)
import numpy as np
import matplotlib.pyplot as plt
from openbox.utils.config_space import ConfigurationSpace, UniformFloatHyperparameter
from openbox.optimizer.parallel_smbo import pSMBO
# Define Configuration Space
config_space = ConfigurationSpace()
x1 = UniformFloatHyperparameter("x1", -5, 10, default_value=0)
x2 = UniformFloatHyperparameter("x2", 0, 15, default_value=0)
config_space.add_hyperparameters([x1, x2])
# Define Objective Function
def branin(config):
config_dict = config.get_dictionary()
x1 = config_dict['x1']
x2 = config_dict['x2']
a = 1.
b = 5.1 / (4. * np.pi**2)
c = 5. / np.pi
r = 6.
s = 10.
t = 1. / (8. * np.pi)
y = a * (x2 - b * x1**2 + c * x1 - r)**2 + s * (1 - t) * np.cos(x1) + s
ret = dict(objs=(y, ))
return ret
# Parallel Evaluation on Local Machine
def townsend(config):
config_dict = config.get_dictionary()
X = np.array([config_dict['x%d' % i] for i in range(2)])
res = dict()
res['objs'] = (-(np.cos(
(X[0] - 0.1) * X[1])**2 + X[0] * np.sin(3 * X[0] + X[1])), )
res['constraints'] = (-(-np.cos(1.5 * X[0] + np.pi) * np.cos(1.5 * X[1]) +
np.sin(1.5 * X[0] + np.pi) * np.sin(1.5 * X[1])), )
return res
townsend_params = {'float': {'x0': (-2.25, 2.5, 0), 'x1': (-2.5, 1.75, 0)}}
townsend_cs = ConfigurationSpace()
townsend_cs.add_hyperparameters([
UniformFloatHyperparameter(name, *para)
for name, para in townsend_params['float'].items()
])
def mishra(config):
config_dict = config.get_dictionary()
X = np.array([config_dict['x%d' % i] for i in range(2)])
x, y = X[0], X[1]
t1 = np.sin(y) * np.exp((1 - np.cos(x))**2)
t2 = np.cos(x) * np.exp((1 - np.sin(y))**2)
t3 = (x - y)**2
result = dict()
result['objs'] = [t1 + t2 + t3]
result['constraints'] = [np.sum((X + 5)**2) - 25]
return result
f1 = (px2 - 5.1 / (4 * np.pi**2) * px1**2 + 5 / np.pi * px1 -
6)**2 + 10 * (1 - 1 / (8 * np.pi)) * np.cos(px1) + 10
f2 = (1 - np.exp(-1 /
(2 * x2))) * (2300 * x1**3 + 1900 * x1**2 + 2092 * x1 +
60) / (100 * x1**3 + 500 * x1**2 + 4 * x1 +
20)
res['objs'] = [f1, f2]
res['constraints'] = []
return res
bc_params = {'float': {'x1': (0, 1, 0.5), 'x2': (0, 1, 0.5)}}
bc_cs = ConfigurationSpace()
bc_cs.add_hyperparameters([
UniformFloatHyperparameter(e, *bc_params['float'][e])
for e in bc_params['float']
])
bc_max_hv = 59.36011874867746
bc_ref_point = [18., 6.]
bo = SMBO(branin_currin,
bc_cs,
advisor_type='mcadvisor',
task_id='mcparego',
num_objs=2,
acq_type='mcparego',
ref_point=bc_ref_point,
max_runs=100,
random_state=2)
bo.run()
c = 5. / np.pi
r = 6.
s = 10.
t = 1. / (8. * np.pi)
ret = a * (x2 - b * x1 ** 2 + c * x1 - r) ** 2 + s * (1 - t) * np.cos(x1) + s
result = dict()
result['objs'] = (ret, )
return result
cs = ConfigurationSpace()
x1 = UniformFloatHyperparameter("x1", -5, 10, default_value=0)
x2 = UniformFloatHyperparameter("x2", 0, 15, default_value=0)
cs.add_hyperparameters([x1, x2])
i = 10
bo = SMBO(branin, cs, advisor_type='default', surrogate_type='gp',
acq_optimizer_type='local_random', initial_runs=3,
task_id='local_random_bo', random_state=i, max_runs=31, time_limit_per_trial=3, logging_dir='logs')
bo.run()
bo2 = SMBO(branin, cs, advisor_type='default', surrogate_type='gp',
acq_optimizer_type='random_scipy', initial_runs=3,
task_id='random_scipy_bo', random_state=i, max_runs=31, time_limit_per_trial=3, logging_dir='logs')
bo2.run()
print(bo.get_incumbent())
print(bo2.get_incumbent())
res['objs'] = [
-(np.cos((X[0] - 0.1) * X[1])**2 + X[0] * np.sin(3 * X[0] + X[1]))
]
res['constraints'] = [
-(-np.cos(1.5 * X[0] + np.pi) * np.cos(1.5 * X[1]) +
np.sin(1.5 * X[0] + np.pi) * np.sin(1.5 * X[1]))
]
return res
# Send task id and config space at register
task_id = time.time()
townsend_params = {'float': {'x1': (-2.25, 2.5, 0), 'x2': (-2.5, 1.75, 0)}}
townsend_cs = ConfigurationSpace()
townsend_cs.add_hyperparameters([
UniformFloatHyperparameter(e, *townsend_params['float'][e])
for e in townsend_params['float']
])
max_runs = 50
# Create remote advisor
config_advisor = RemoteAdvisor(config_space=townsend_cs,
server_ip='127.0.0.1',
port=11425,
email='*****@*****.**',
password='******',
num_constraints=1,
max_runs=max_runs,
task_name="task_test",
task_id=task_id)
# Simulate max_runs iterations
def get_configspace():
cs = ConfigurationSpace()
x1 = UniformFloatHyperparameter("x1", -5, 10, default_value=0)
x2 = UniformFloatHyperparameter("x2", 0, 15, default_value=0)
cs.add_hyperparameters([x1, x2])
return cs