Secure your code as it's written. Use Snyk Code to scan source code in minutes - no build needed - and fix issues immediately.
def __init__(
self,
layer_size,
activation,
kernel_initializer,
regularization=None,
dropout_rate=0,
batch_normalization=None,
):
self.layer_size = layer_size
self.activation = activations.get(activation)
self.kernel_initializer = initializers.get(kernel_initializer)
self.regularizer = regularizers.get(regularization)
self.dropout_rate = dropout_rate
self.batch_normalization = batch_normalization
super(FNN, self).__init__()
def __init__(
self,
layer_size_low_fidelity,
layer_size_high_fidelity,
activation,
kernel_initializer,
regularization=None,
residue=False,
):
self.layer_size_lo = layer_size_low_fidelity
self.layer_size_hi = layer_size_high_fidelity
self.activation = activations.get(activation)
self.kernel_initializer = initializers.get(kernel_initializer)
self.regularizer = regularizers.get(regularization)
self.residue = residue
super(MfNN, self).__init__()
def __init__(
self,
input_size,
output_size,
num_neurons,
num_blocks,
activation,
kernel_initializer,
regularization=None,
):
self.input_size = input_size
self.output_size = output_size
self.num_neurons = num_neurons
self.num_blocks = num_blocks
self.activation = activations.get(activation)
self.kernel_initializer = initializers.get(kernel_initializer)
self.regularizer = regularizers.get(regularization)
super(ResNet, self).__init__()
activation=self.activation,
regularizer=self.regularizer,
)
y_hi_nl = self.dense(
y, self.layer_size_hi[-1], use_bias=False, regularizer=self.regularizer
)
# Linear + nonlinear
if not self.residue:
alpha = tf.Variable(0, dtype=config.real(tf))
alpha = activations.get("tanh")(alpha)
self.y_hi = y_hi_l + alpha * y_hi_nl
else:
alpha1 = tf.Variable(0, dtype=config.real(tf))
alpha1 = activations.get("tanh")(alpha1)
alpha2 = tf.Variable(0, dtype=config.real(tf))
alpha2 = activations.get("tanh")(alpha2)
self.y_hi = self.y_lo + 0.1 * (alpha1 * y_hi_l + alpha2 * y_hi_nl)
self.target_lo = tf.placeholder(config.real(tf), [None, self.layer_size_lo[-1]])
self.target_hi = tf.placeholder(config.real(tf), [None, self.layer_size_hi[-1]])
y,
self.layer_size_hi[i],
activation=self.activation,
regularizer=self.regularizer,
)
y_hi_nl = self.dense(
y, self.layer_size_hi[-1], use_bias=False, regularizer=self.regularizer
)
# Linear + nonlinear
if not self.residue:
alpha = tf.Variable(0, dtype=config.real(tf))
alpha = activations.get("tanh")(alpha)
self.y_hi = y_hi_l + alpha * y_hi_nl
else:
alpha1 = tf.Variable(0, dtype=config.real(tf))
alpha1 = activations.get("tanh")(alpha1)
alpha2 = tf.Variable(0, dtype=config.real(tf))
alpha2 = activations.get("tanh")(alpha2)
self.y_hi = self.y_lo + 0.1 * (alpha1 * y_hi_l + alpha2 * y_hi_nl)
self.target_lo = tf.placeholder(config.real(tf), [None, self.layer_size_lo[-1]])
self.target_hi = tf.placeholder(config.real(tf), [None, self.layer_size_hi[-1]])
# Nonlinear
y = X_hi
for i in range(len(self.layer_size_hi) - 1):
y = self.dense(
y,
self.layer_size_hi[i],
activation=self.activation,
regularizer=self.regularizer,
)
y_hi_nl = self.dense(
y, self.layer_size_hi[-1], use_bias=False, regularizer=self.regularizer
)
# Linear + nonlinear
if not self.residue:
alpha = tf.Variable(0, dtype=config.real(tf))
alpha = activations.get("tanh")(alpha)
self.y_hi = y_hi_l + alpha * y_hi_nl
else:
alpha1 = tf.Variable(0, dtype=config.real(tf))
alpha1 = activations.get("tanh")(alpha1)
alpha2 = tf.Variable(0, dtype=config.real(tf))
alpha2 = activations.get("tanh")(alpha2)
self.y_hi = self.y_lo + 0.1 * (alpha1 * y_hi_l + alpha2 * y_hi_nl)
self.target_lo = tf.placeholder(config.real(tf), [None, self.layer_size_lo[-1]])
self.target_hi = tf.placeholder(config.real(tf), [None, self.layer_size_hi[-1]])
layer_size_function,
layer_size_location,
activation,
kernel_initializer,
regularization=None,
use_bias=True,
stacked=False,
):
if layer_size_function[-1] != layer_size_location[-1]:
raise ValueError(
"Output sizes of function NN and location NN do not match."
)
self.layer_size_func = layer_size_function
self.layer_size_loc = layer_size_location
self.activation = activations.get(activation)
self.kernel_initializer = initializers.get(kernel_initializer)
if stacked:
self.kernel_initializer_stacked = initializers.get(
kernel_initializer + "stacked"
)
self.regularizer = regularizers.get(regularization)
self.use_bias = use_bias
self.stacked = stacked
super(OpNN, self).__init__()