def __init__(self):
hp_space = [
HyperParameter.float_param('alpha', (0.0, 10.0)),
HyperParameter.float_param('binarize', (0.0, 1.0))
]
initializer = sklearn.naive_bayes.BernoulliNB
super().__init__(hp_space, initializer)
def __init__(self):
hp_space = [
HyperParameter.float_param('reg_param', (0., 1.)),
HyperParameter.float_param('tol', (1e-5, 1e-3))
]
initializer = sklearn.discriminant_analysis.QuadraticDiscriminantAnalysis
super().__init__(hp_space, initializer)
def __init__(self):
hp_space = [
HyperParameter.float_param('learning_rate', (1e-6, 0.15)),
HyperParameter.categorical_param('use_locking', (True, False))
]
model_initializer = tf.train.GradientDescentOptimizer
super().__init__(hp_space, model_initializer)
def __init__(self):
hp_space = [
HyperParameter.int_param('n_estimators', (50, 500)),
HyperParameter.float_param('learning_rate', (0.1, 2.)),
HyperParameter.categorical_param('algorithm', ('SAMME', 'SAMME.R')),
]
initializer = sklearn.ensemble.AdaBoostClassifier
super().__init__(hp_space, initializer)
def __init__(self):
hp_space = [
HyperParameter.float_param('alpha', (1e-2, 100.0)),
HyperParameter.categorical_param('fit_prior', (True, False))
# HyperParameter.float_param('binarize', (0.0, 1.0))
]
initializer = sklearn.naive_bayes.BernoulliNB
super().__init__(hp_space, initializer)
def __init__(self):
hp_space = [
HyperParameter.int_param('n_neighbors', (1, 100)),
HyperParameter.categorical_param('weights', ('uniform', 'distance')),
HyperParameter.categorical_param('p', (1, 2))
]
initializer = sklearn.neighbors.KNeighborsClassifier
super().__init__(hp_space, initializer)
def __init__(self):
hp_space = [
HyperParameter.int_param('n_estimators', (5, 100)),
HyperParameter.float_param('max_samples', (0.0, 1.0)),
# HyperParameter.float_param('max_features', (0.0, 1.0))
]
initializer = sklearn.ensemble.BaggingClassifier
super().__init__(hp_space, initializer)
def __init__(self):
hp_space = [
HyperParameter.float_param('learning_rate', (1e-6, 0.15)),
HyperParameter.float_param('momentum', (0., 1.)),
HyperParameter.categorical_param('use_locking', (True, False)),
HyperParameter.categorical_param('use_nesterov', (True, False))
]
model_initializer = tf.train.MomentumOptimizer
super().__init__(hp_space, model_initializer)
def __init__(self):
hp_space = [
HyperParameter.float_param('learning_rate', (1e-6, 0.15)),
HyperParameter.float_param('initial_accumulator_value',
(1e-6, 2.)),
HyperParameter.categorical_param('use_locking', (True, False))
]
model_initializer = tf.train.AdagradOptimizer
super().__init__(hp_space, model_initializer)
def __init__(self):
hp_space = [
HyperParameter.float_param('learning_rate', (1e-6, 0.15)),
HyperParameter.float_param('l1_regularization_strength', (0., 2.)),
HyperParameter.float_param('l2_regularization_strength', (0., 2.)),
HyperParameter.categorical_param('use_locking', (True, False))
]
model_initializer = tf.train.ProximalGradientDescentOptimizer
super().__init__(hp_space, model_initializer)
def __init__(self):
hp_space = [
HyperParameter.float_param('alpha', (1e-2, 1e2)),
HyperParameter.int_param('max_iter', (1000, 10000)),
HyperParameter.float_param('tol', (1e-4, 1e-2)),
HyperParameter.categorical_param('solver', ('svd', 'cholesky', 'lsqr', 'sparse_cg', 'sag', 'saga'))
]
initializer = sklearn.linear_model.RidgeClassifier
super().__init__(hp_space, initializer)
def __init__(self):
hp_space = [
HyperParameter.float_param('learning_rate', (1e-6, 0.15)),
HyperParameter.float_param('beta1', (0., 1. - 1e-6)),
HyperParameter.float_param('beta2', (0., 1. - 1e-6)),
HyperParameter.float_param('epsilon', (1e-6, 0.1)),
HyperParameter.categorical_param('use_locking', (True, False))
]
model_initializer = tf.train.AdamOptimizer
super().__init__(hp_space, model_initializer)
def __init__(self):
hp_space = [
HyperParameter.float_param('C', (0.1, 10)),
HyperParameter.int_param('max_iter', (1000, 10000)),
HyperParameter.float_param('tol', (1e-4, 1e-2)),
HyperParameter.categorical_param('loss',
('hinge', 'squared_hinge'))
]
initializer = sklearn.linear_model.PassiveAggressiveClassifier
super().__init__(hp_space, initializer)
def __init__(self):
hp_space = [
HyperParameter.categorical_param('penalty', (None, 'l2', 'l1', 'elasticnet')),
HyperParameter.float_param('alpha', (1e-5, 1e-3)),
HyperParameter.int_param('max_iter', (1000, 10000)),
HyperParameter.float_param('tol', (1e-4, 1e-2)),
HyperParameter.float_param('eta0', (0.1, 10))
]
initializer = sklearn.linear_model.Perceptron
super().__init__(hp_space, initializer)
def __init__(self):
hp_space = [
HyperParameter.float_param('C', (1e-5, 10)),
HyperParameter.categorical_param('fit_intercept', (True,)),
HyperParameter.float_param('tol', (1e-5, 1e-1)),
HyperParameter.categorical_param('loss', ('hinge', 'squared_hinge')),
HyperParameter.categorical_param('average', (False, True))
]
initializer = sklearn.linear_model.PassiveAggressiveClassifier
super().__init__(hp_space, initializer)
def __init__(self):
hp_space = [
HyperParameter.float_param('radius', (1e-2, 1e3)),
HyperParameter.categorical_param('weights', ('uniform', 'distance')),
HyperParameter.categorical_param('algorithm', ('ball_tree', 'kd_tree', 'brute')),
HyperParameter.int_param('leaf_size', (3, 100)),
HyperParameter.int_param('p', (1, 10))
]
initializer = sklearn.neighbors.RadiusNeighborsClassifier
super().__init__(hp_space, initializer)
def __init__(self):
hp_space = [
HyperParameter.categorical_param('penalty', ('l1', 'l2')),
HyperParameter.categorical_param('loss', ('hinge', 'squared_hinge')),
HyperParameter.categorical_param('dual', (True, False)),
HyperParameter.float_param('tol', (1e-6, 1e-1)),
HyperParameter.float_param('C', (0.03125, 327686))
]
initializer = sklearn.svm.LinearSVC
super().__init__(hp_space, initializer)
def __init__(self):
hp_space = [
HyperParameter.float_param('learning_rate', (1e-6, 0.15)),
HyperParameter.float_param('rho', (1e-6, 1.)), #recheck needed
HyperParameter.float_param(
'epsilon',
(1e-6, 0.1)), #wider range is allowed but not suggested
HyperParameter.categorical_param('use_locking', (True, False))
]
model_initializer = tf.train.AdadeltaOptimizer
super().__init__(hp_space, model_initializer)
def __init__(self):
hp_space = [
HyperParameter.float_param('tol', (1e-6, 1e-1)),
HyperParameter.float_param('C', (1e-2, 1e2)),
HyperParameter.categorical_param('solver',
('newton-cg', 'lbfgs', 'sag')),
HyperParameter.int_param('max_iter', (100, 1000)),
HyperParameter.categorical_param('multi_class',
('ovr', 'multinomial'))
]
initializer = sklearn.linear_model.LogisticRegression
super().__init__(hp_space, initializer)
def __init__(self):
hp_space = [
HyperParameter.categorical_param('criterion', ('gini', 'entropy')),
HyperParameter.int_param('max_depth', (1, 40)),
HyperParameter.int_param('min_samples_split', (2, 20)),
HyperParameter.int_param('min_samples_leaf', (1, 20)),
HyperParameter.categorical_param('max_features', ('sqrt', 'log2', None)),
# HyperParameter.int_param('max_leaf_nodes', (2, 100)),
# HyperParameter.float_param('min_impurity_decrease', (0., 100.)),
]
model_initializer = sklearn.tree.DecisionTreeClassifier
super().__init__(hp_space, model_initializer)
def __init__(self):
hp_space = [
HyperParameter.float_param('alpha', (1e-2, 100.0))
]
initializer = sklearn.naive_bayes.MultinomialNB
super().__init__(hp_space, initializer)
def __init__(self):
hp_space = [
HyperParameter.int_param('max_iter_predict', (10, 1000))
]
initializer = sklearn.gaussian_process.GaussianProcessClassifier
super().__init__(hp_space, initializer)
def __init__(self):
hp_space = [
HyperParameter.float_param('var_smoothing', (1e-10, 1e-8))
]
initializer = sklearn.naive_bayes.GaussianNB
super().__init__(hp_space, initializer)
def __init__(self):
hp_space = [
HyperParameter.int_param('n_estimators', (5, 1000)),
HyperParameter.categorical_param('criterion', ('gini', 'entropy')),
HyperParameter.int_param('max_depth', (2, 40)),
HyperParameter.int_param('min_samples_split', (1, 100)),
HyperParameter.int_param('min_samples_leaf', (1, 100)),
HyperParameter.categorical_param('max_features',
('sqrt', 'log2', None)),
HyperParameter.int_param('max_leaf_nodes', (-1, 100)),
HyperParameter.float_param('min_impurity_decrease', (0.0, 100.0))
]
initializer = sklearn.ensemble.ExtraTreesClassifier
super().__init__(hp_space, initializer)
def __init__(self):
hp_space = [
HyperParameter.categorical_param('loss',
('hinge', 'log', 'modified_huber',
'squared_hinge', 'perceptron')),
HyperParameter.categorical_param(
'penalty', ('none', 'l2', 'l1', 'elasticnet')),
HyperParameter.float_param('alpha', (1e-5, 1e-3)),
HyperParameter.float_param('l1_ratio', (0.0, 1.0)),
HyperParameter.int_param('max_iter', (1000, 10000)),
HyperParameter.float_param('tol', (1e-4, 1e-2)),
HyperParameter.categorical_param(
'learning_rate',
('constant', 'optimal', 'invscaling', 'adaptive')),
HyperParameter.float_param('eta0', (0.0, 10.0)),
HyperParameter.float_param('power_t', (0.05, 1))
]
initializer = sklearn.linear_model.SGDClassifier
super().__init__(hp_space, initializer)
def __init__(self):
hp_space = [
HyperParameter.float_param('nu', (5e-3, 1)),
HyperParameter.categorical_param(
'kernel', ('linear', 'poly', 'rbf', 'sigmoid')),
HyperParameter.int_param('degree', (1, 30)),
HyperParameter.float_param('gamma', (1e-5, 10)),
HyperParameter.float_param('coef0', (0., 100.)),
HyperParameter.categorical_param('shrinking', (True, False)),
HyperParameter.float_param('tol', (1e-5, 1e-1))
]
model_initializer = sklearn.svm.NuSVC
super().__init__(hp_space, model_initializer)
def __init__(self):
hp_space = [
HyperParameter.float_param('learning_rate', (1e-6, 0.15)),
HyperParameter.float_param('learning_rate_power', (-2., -1e-6)),
HyperParameter.float_param('initial_accumulator_value', (0., 2.)),
HyperParameter.float_param('l1_regularization_strength', (0., 2.)),
HyperParameter.float_param('l2_regularization_strength', (0., 2.)),
HyperParameter.categorical_param('use_locking', (True, False)),
HyperParameter.float_param('l2_shrinkage_regularization_strength',
(0., 2.)),
]
model_initializer = tf.train.FtrlOptimizer
super().__init__(hp_space, model_initializer)
def __init__(self):
hp_space = [
HyperParameter.float_param('learning_rate', (1e-6, 0.15)),
HyperParameter.float_param('decay', (0., 1.)),
HyperParameter.float_param('momentum', (0., 1.)),
HyperParameter.float_param('epsilon', (1e-6, 0.1)),
HyperParameter.categorical_param('use_locking', (True, False)),
HyperParameter.categorical_param('centered', (True, False)),
]
model_initializer = tf.trainRMSPropOptimizer
super().__init__(hp_space, model_initializer)
def __init__(self):
hp_space = [
HyperParameter.categorical_param('n_estimators', (100,)),
HyperParameter.categorical_param('criterion', ('gini', 'entropy')),
HyperParameter.int_param('min_samples_split', (2, 20)),
HyperParameter.int_param('min_samples_leaf', (1, 20)),
HyperParameter.float_param('max_features', (0., 1.)),
HyperParameter.categorical_param('bootstrap', (True, False)),
]
initializer = sklearn.ensemble.RandomForestClassifier
super().__init__(hp_space, initializer)
def __init__(self):
hp_space = [
HyperParameter.float_param('learning_rate', (1e-6, 0.15)),
HyperParameter.int_param('global_step', (0, 1e9)),
HyperParameter.float_param(
'initial_gradient_squared_accumulator_value', (1e-6, 1.)),
HyperParameter.float_param('l1_regularization_strength', (0., 1.)),
HyperParameter.float_param('l2_regularization_strength', (0., 1.)),
HyperParameter.categorical_param('use_locking', (True, False))
]
model_initializer = tf.train.AdagradDAOptimizer
super().__init__(hp_space, model_initializer)