def __init__(self,
embedding_matrix=None,
backwards=True,
dropout=0.4,
optimizer="rmsprop",
lstm_out_width=20,
lstm_pool_size=128,
learn_rate=1.0,
verbose=0,
batch_size=32,
epochs=35,
shuffle=False,
class_weight=30.0):
"""Initialize the LSTM pool model."""
super(LSTMPoolClassifier, self).__init__()
self.embedding_matrix = embedding_matrix
self.backwards = backwards
self.dropout = dropout
self.optimizer = optimizer
self.lstm_out_width = lstm_out_width
self.learn_rate = learn_rate
self.verbose = verbose
self.batch_size = batch_size
self.epochs = epochs
self.shuffle = shuffle
self.class_weight = _set_class_weight(class_weight)
self.lstm_pool_size = lstm_pool_size
self._model = None
self.sequence_length = None
def __init__(self, C=1.0, class_weight=1.0, random_state=None, n_jobs=1):
"""Initialize the SKLearn Naive Bayes model.
Arguments:
----------
C: float
Parameter inverse to the regularization strength of the model.
class_weight: float
Class weight of the inclusions.
random_state: int, RandomState
Random state for the model.
n_jobs: int
Number of CPU cores used.
"""
super(LogisticModel, self).__init__()
self.C = C
self.class_weight = class_weight
self.n_jobs = n_jobs
self._model = LogisticRegression(
solver="liblinear",
C=C,
class_weight=_set_class_weight(class_weight),
n_jobs=n_jobs,
random_state=random_state)
logging.debug(self._model)
def __init__(self,
gamma="auto",
class_weight=0.249,
C=15.4,
kernel="linear",
random_state=None):
"""Initialize the SKLearn SVM model.
Arguments
---------
gamma: str
Gamma parameter of the SVM model.
class_weight: float
class_weight of the inclusions.
C: float
C parameter of the SVM model.
kernel: str
SVM kernel type.
random_state: int, RandomState
State of the RNG.
"""
super(SVMModel, self).__init__()
self.gamma = gamma
self.class_weight = class_weight
self.C = C
self.kernel = kernel
self._random_state = random_state
self._model = SVC(kernel=kernel,
C=C,
class_weight=_set_class_weight(class_weight),
random_state=random_state,
gamma=gamma,
probability=True)
def fit(self, X, y):
# check is tensorflow is available
_check_tensorflow()
sequence_length = X.shape[1]
if self._model is None or sequence_length != self.sequence_length:
self.sequence_length = sequence_length
keras_model = _create_lstm_base_model(
embedding_matrix=self.embedding_matrix,
backwards=self.backwards,
dropout=self.dropout,
optimizer=self.optimizer,
max_sequence_length=self.sequence_length,
lstm_out_width=self.lstm_out_width,
dense_width=self.dense_width,
learn_rate=self.learn_rate,
verbose=self.verbose)
print(keras_model)
self._model = KerasClassifier(keras_model, verbose=self.verbose)
self._model.fit(
X,
y,
batch_size=self.batch_size,
epochs=self.epochs,
shuffle=self.shuffle,
class_weight=_set_class_weight(self.class_weight),
verbose=self.verbose)
def lstm_fit_defaults(settings, verbose=1):
""" Set the fit defaults and merge them with custom settings. """
# arguments to pass to the fit
fit_kwargs = {}
fit_kwargs['batch_size'] = 32
fit_kwargs['epochs'] = 10
fit_kwargs['verbose'] = verbose
fit_kwargs['shuffle'] = False
fit_kwargs['class_weight_inc'] = 30.0
settings.fit_kwargs = _unsafe_dict_update(fit_kwargs, settings.fit_param)
_set_class_weight(fit_kwargs.pop('class_weight_inc'), fit_kwargs)
return settings.fit_kwargs
def __init__(self, n_estimators=100, max_features=10, class_weight=1.0,
random_state=None):
"""Initialize the SKLearn Random Forest model.
Arguments
---------
n_estimators: int
Number of estimators.
max_features: int
Number of features in the model.
class_weight: float
Class weight of the inclusions.
random_state: int, RandomState
Set the random state of the RNG.
"""
super(RFModel, self).__init__()
self.n_estimators = int(n_estimators)
self.max_features = int(max_features)
self.class_weight = class_weight
self.random_state = random_state
self._model = RandomForestClassifier(
n_estimators=self.n_estimators, max_features=self.max_features,
class_weight=_set_class_weight(class_weight),
random_state=random_state)
def __init__(self,
embedding_matrix=None,
backwards=True,
dropout=0.4,
optimizer="rmsprop",
lstm_out_width=20,
lstm_pool_size=128,
learn_rate=1.0,
verbose=0,
batch_size=32,
epochs=35,
shuffle=False,
class_weight=30.0):
"""Initialize the LSTM pool model.
Arguments
---------
embedding_matrix: np.array
Embedding matrix to use with LSTM model.
backwards: bool
Whether to have a forward or backward LSTM.
dropout: float
Value in [0, 1.0) that gives the dropout and recurrent
dropout rate for the LSTM model.
optimizer: str
Optimizer to use.
lstm_out_width: int
Output width of the LSTM.
lstm_pool_size: int
Size of the pool, must be a divisor of max_sequence_length.
learn_rate: float
Learn rate multiplier of default learning rate.
verbose: int
Verbosity.
batch_size: int
Size of the batch size for the LSTM model.
epochs: int
Number of epochs to train the LSTM model.
shuffle: bool
Whether to shuffle the data before starting to train.
class_weight: float
Class weight for the included papers.
"""
super(LSTMPoolModel, self).__init__()
self.embedding_matrix = embedding_matrix
self.backwards = backwards
self.dropout = dropout
self.optimizer = optimizer
self.lstm_out_width = lstm_out_width
self.learn_rate = learn_rate
self.verbose = verbose
self.batch_size = batch_size
self.epochs = epochs
self.shuffle = shuffle
self.class_weight = _set_class_weight(class_weight)
self.lstm_pool_size = lstm_pool_size
self._model = None
self.sequence_length = None
def fit(self, X, y):
if scipy.sparse.issparse(X):
X = X.toarray()
if self._model is None or X.shape[1] != self.input_dim:
self.input_dim = X.shape[1]
keras_model = _create_dense_nn_model(
self.input_dim, self.dense_width, self.optimizer,
self.learn_rate, self.regularization, self.verbose)
self._model = KerasClassifier(keras_model, verbose=self.verbose)
self._model.fit(X, y, batch_size=self.batch_size, epochs=self.epochs,
shuffle=self.shuffle, verbose=self.verbose,
class_weight=_set_class_weight(self.class_weight))
def __init__(self, C=1.0, class_weight=1.0, random_state=None, n_jobs=1):
super(LogisticClassifier, self).__init__()
self.C = C
self.class_weight = class_weight
self.n_jobs = n_jobs
self._model = LogisticRegression(
solver="liblinear",
C=C,
class_weight=_set_class_weight(class_weight),
n_jobs=n_jobs,
random_state=random_state)
logging.debug(self._model)
def __init__(self,
n_estimators=100,
max_features=10,
class_weight=1.0,
random_state=None):
super(RandomForestClassifier, self).__init__()
self.n_estimators = int(n_estimators)
self.max_features = int(max_features)
self.class_weight = class_weight
self._random_state = random_state
self._model = SKRandomForestClassifier(
n_estimators=self.n_estimators,
max_features=self.max_features,
class_weight=_set_class_weight(class_weight),
random_state=random_state)
def __init__(self, C=1.0, class_weight=1.0, n_jobs=1):
"""Initialize the SKLearn Naive Bayes model.
Arguments:
----------
alpha: float
Parameter to set the regularization strength of the model.
"""
super(LogisticModel, self).__init__()
self.C = C
self.class_weight = class_weight
self.n_jobs = n_jobs
self._model = LogisticRegression(
solver="liblinear", C=C,
class_weight=_set_class_weight(class_weight),
n_jobs=n_jobs)
logging.debug(self._model)
def __init__(self,
gamma="auto",
class_weight=0.249,
C=15.4,
kernel="linear",
random_state=None):
super(SVMClassifier, self).__init__()
self.gamma = gamma
self.class_weight = class_weight
self.C = C
self.kernel = kernel
self._random_state = random_state
self._model = SVC(kernel=kernel,
C=C,
class_weight=_set_class_weight(class_weight),
random_state=random_state,
gamma=gamma,
probability=True)
def fit_kwargs(self):
fit_kwargs = self.fit_param()
class_weight_inc = fit_kwargs.pop('class_weight_inc', None)
if class_weight_inc is not None:
fit_kwargs['class_weight'] = _set_class_weight(class_weight_inc)
return fit_kwargs
