def weight_variable(shape, stddev=0.05, name='weight'):
# initial=tf.truncated_normal(shape=shape,dtype=tf.float32,stddev=stddev)
initial = variance_scaling_initializer(factor=1.0,
mode='FAN_AVG',
uniform=True)
# return tf.Variable(initial,dtype=tf.float32,name=name)
return tf.Variable(initial(shape=shape), name=name)
def conv2d(input_,
output_dim,
k_h=3,
k_w=3,
d_h=2,
d_w=2,
padding='SAME',
name="conv2d",
with_w=False):
with tf.variable_scope(name):
w = tf.get_variable(
'w', [k_h, k_w, input_.get_shape()[-1], output_dim],
initializer=variance_scaling_initializer())
if padding == 'Other':
padding = 'VALID'
input_ = tf.pad(input_, [[0, 0], [3, 3], [3, 3], [0, 0]],
"CONSTANT")
elif padding == 'VALID':
padding = 'VALID'
conv = tf.nn.conv2d(input_,
w,
strides=[1, d_h, d_w, 1],
padding=padding)
biases = tf.get_variable('biases', [output_dim],
initializer=tf.constant_initializer(0.0))
conv = tf.reshape(tf.nn.bias_add(conv, biases), conv.get_shape())
if with_w:
return conv, w, biases
else:
return conv
def create_cond_critic_proj(cls, xinput, input_clusters, var_scope,
critic_layers, clusters_no, reuse=None):
"""
Class method that instantiates a Critic and creates a conditional
critic with the original projection conditioning method.
Parameters
----------
xinput : Tensor
Tensor containing the input cells.
input_clusters : Tensor
Tensor containing the corresponding cluster indexes of the input cells.
var_scope : str
Variable scope used for the created tensors.
critic_layers : list
List of integers corresponding to the number of neurons of each
layer of the critic.
clusters_no : int
Number of clusters.
reuse : Boolean
Whether to reuse the already existing Tensors.
Default is None.
Returns
-------
A Creator object with the defined architecture.
"""
with tf.variable_scope(var_scope, reuse=reuse):
for i_lay, output_size in enumerate(critic_layers):
with tf.variable_scope("layers_" + str(i_lay + 1)):
xinput = layers.relu(
xinput,
output_size,
weights_initializer=layers.variance_scaling_initializer(mode="FAN_AVG"),
biases_initializer=tf.zeros_initializer())
with tf.variable_scope("layers_" + 'proj'):
proj_weights_m = tf.get_variable(
"proj_weights_m",
[clusters_no, critic_layers[-1], 1],
dtype=tf.float32, initializer=layers.xavier_initializer())
proj_weights = tf.nn.embedding_lookup(proj_weights_m,
input_clusters)
output_proj = tf.einsum('ij,ijk->ik', xinput, proj_weights)
with tf.variable_scope("layers_" + 'output'):
output = layers.linear(
xinput, 1,
weights_initializer=layers.xavier_initializer(),
biases_initializer=tf.zeros_initializer())
dist = tf.add(output_proj, output)
return cls(xinput, dist, var_scope, critic_layers,
input_clusters=input_clusters,
clusters_no=clusters_no, reuse=reuse)
def de_conv(input_, output_shape,
k_h=3, k_w=3, d_h=2, d_w=2, stddev=0.02,
name="deconv2d", with_w=False):
with tf.variable_scope(name):
# filter : [height, width, output_channels, in_channels]
w = tf.get_variable('w', [k_h, k_w, output_shape[-1], input_.get_shape()[-1]],
initializer=variance_scaling_initializer())
try:
deconv = tf.nn.conv2d_transpose(input_, w, output_shape=output_shape,
strides=[1, d_h, d_w, 1])
# Support for verisons of TensorFlow before 0.7.0
except AttributeError:
deconv = tf.nn.deconv2d(input_, w, output_shape=output_shape,
strides=[1, d_h, d_w, 1])
biases = tf.get_variable('biases', [output_shape[-1]], initializer=tf.constant_initializer(0.0))
deconv = tf.reshape(tf.nn.bias_add(deconv, biases), deconv.get_shape())
if with_w:
return deconv, w, biases
else:
return deconv
def de_conv2d(input_,
output_shape,
k_h=3,
k_w=3,
d_h=2,
d_w=2,
stddev=2e-2,
name='deconv2d',
with_w=False,
initializer=variance_scaling_initializer()):
with tensorflow.variable_scope(name):
w = tensorflow.get_variable(
'w', [k_h, k_w, output_shape[-1],
input_.get_shape()[-1]],
initializer=initializer)
deconv = tensorflow.nn.conv2d_transpose(input_,
w,
output_shape=output_shape,
strides=[1, d_h, d_w, 1])
b = tensorflow.get_variable(
'b', [output_shape[-1]],
initializer=tensorflow.constant_initializer(0.0))
deconv = tensorflow.reshape(tensorflow.nn.bias_add(deconv, b),
deconv.get_shape())
if with_w:
return deconv, w, b
else:
return deconv
def fully_connect(input_, output_size, scope=None, with_w=False):
shape = input_.get_shape().as_list()
with tf.variable_scope(scope or "Linear"):
matrix = tf.get_variable("Matrix", [shape[1], output_size], tf.float32,
variance_scaling_initializer())
bias = tf.get_variable("bias", [output_size],
initializer=tf.constant_initializer(0.0))
if with_w:
return tf.matmul(input_, matrix) + bias, matrix, bias
else:
return tf.matmul(input_, matrix) + bias
def conv2d(input_, output_dim, k_h=3, k_w=3, d_h=2, d_w=2, name='conv2d'):
with tensorflow.variable_scope(name):
w = tensorflow.get_variable(
'w', [k_h, k_w, input_.get_shape()[-1], output_dim],
initializer=variance_scaling_initializer())
conv = tensorflow.nn.conv2d(input_,
w,
strides=[1, d_h, d_w, 1],
padding='SAME')
b = tensorflow.get_variable(
'b', [output_dim],
initializer=tensorflow.constant_initializer(0.0))
conv = tensorflow.reshape(tensorflow.nn.bias_add(conv, b),
conv.get_shape())
return conv, w
def fully_connected(input_,
output_size,
scope=None,
with_w=False,
initializer=variance_scaling_initializer()):
shape = input_.get_shape().as_list()
with tensorflow.variable_scope(scope or 'Linear'):
matrix = tensorflow.get_variable('Matrix', [shape[1], output_size],
tensorflow.float32,
initializer=initializer)
b = tensorflow.get_variable(
'b', [output_size],
initializer=tensorflow.constant_initializer(0.0))
if with_w:
return tensorflow.matmul(input_, matrix) + b, matrix, b
else:
return tensorflow.matmul(input_, matrix) + b
def conv2d(input_, output_dim, k_h=3, k_w=3, d_h=2, d_w=2, name="conv2d"):
with tf.variable_scope(name):
w = tf.get_variable(
'w', [k_h, k_w, input_.get_shape()[-1], output_dim],
initializer=variance_scaling_initializer(
)) #方差缩放初始化,[3,3,in_channel,out_channel]
conv = tf.nn.conv2d(input_,
w,
strides=[1, d_h, d_w, 1],
padding='SAME')
biases = tf.get_variable('biases', [output_dim],
initializer=tf.constant_initializer(0.0))
conv = tf.reshape(tf.nn.bias_add(conv, biases), conv.get_shape())
return conv, w
def create_critic(cls, xinput, var_scope,
critic_layers, reuse=None):
"""
Class method that instantiates a Critic and creates a
non-conditional critic.
Parameters
----------
xinput : Tensor
Tensor containing the input cells.
var_scope : str
Variable scope used for the created tensors.
critic_layers : list
List of integers corresponding to the number of neurons of each
layer of the critic.
reuse : Boolean
Whether to reuse the already existing Tensors.
Default is None.
Returns
-------
A Creator object with the defined architecture.
"""
with tf.variable_scope(var_scope, reuse=reuse):
for i_lay, output_size in enumerate(critic_layers):
with tf.variable_scope("layers_" + str(i_lay + 1)):
xinput = layers.relu(
xinput, output_size,
weights_initializer=layers.variance_scaling_initializer(mode="FAN_AVG"),
biases_initializer=tf.zeros_initializer())
with tf.variable_scope("layers_" + 'output'):
output = layers.linear(
xinput, 1,
weights_initializer=layers.xavier_initializer(),
biases_initializer=tf.zeros_initializer())
return cls(xinput, output, var_scope, critic_layers, reuse=reuse)
def de_conv(input_,
output_shape,
k_h=3,
k_w=3,
d_h=2,
d_w=2,
stddev=0.02,
name="deconv2d",
with_w=False,
initializer=variance_scaling_initializer()):
with tf.variable_scope(name):
# filter : [height, width, output_channels, in_channels]
w = tf.get_variable(
'w', [k_h, k_w, output_shape[-1],
input_.get_shape()[-1]],
initializer=initializer) #[3,3,128,138]
try:
deconv = tf.nn.conv2d_transpose(input_,
w,
output_shape=output_shape,
strides=[1, d_h, d_w, 1])
# Support for verisons of TensorFlow before 0.7.0
except AttributeError:
deconv = tf.nn.deconv2d(input_,
w,
output_shape=output_shape,
strides=[1, d_h, d_w, 1])
##第二个参数filter：卷积核，它要求是一个Tensor，具有[filter_height, filter_width, out_channels, in_channels]这样的shape，
# 具体含义是[卷积核的高度，卷积核的宽度，卷积核个数，图像通道数]
biases = tf.get_variable('biases', [output_shape[-1]],
initializer=tf.constant_initializer(0.0))
deconv = tf.reshape(tf.nn.bias_add(deconv, biases), deconv.get_shape())
if with_w:
return deconv, w, biases
else:
return deconv
def create_cond_generator(cls, z_input, batch_size, latent_dim,
output_cells_dim, var_scope, gen_layers,
output_lsn, gen_cond_type, clusters_ratios,
is_training, clusters_no=None,
input_clusters=None, reuse=None):
"""
Class method that instantiates a Generator and creates a
conditional generator.
Parameters
----------
z_input : Tensor
Tensor containing the noise used as input by the generator.
batch_size : int
Batch size used during the training.
latent_dim : int
Dimension of the latent space used from which the input noise
of the generator is sampled.
output_cells_dim : int
Dimension of the output cells (i.e. the number of genes).
var_scope : str
Variable scope used for the created tensors.
gen_layers : list
List of integers corresponding to the number of neurons of
each layer of the generator.
output_lsn : int, None
Parameter of the LSN layer at the output of the critic
(i.e. total number of counts per generated cell).
gen_cond_type : str
conditional normalization layers used in the generator, can be
either "batchnorm" or "layernorm". If anything else, it won't be
added in the model (which means no conditional generation).
clusters_ratios : Tensor
Placeholder containing the list of cluster ratios of the input data.
is_training : Tensor
Boolean placeholder encoding for whether we're in training or
inference mode (for the batch normalization).
clusters_no : int
Number of clusters.
Default is None.
input_clusters : Tensor
Placeholders for the cluster indexes that should be used for
conditional generation.
Default is None.
reuse : Boolean
Whether to reuse the already existing Tensors.
Default is None.
Returns
-------
A conditional Generator object with the defined architecture.
"""
with tf.variable_scope(var_scope, reuse=reuse):
for i_lay, size in enumerate(gen_layers):
with tf.variable_scope("generator_layers_" + str(i_lay + 1)):
z_input = layers.linear(
z_input,
size,
weights_initializer=layers.xavier_initializer(),
biases_initializer=None)
if i_lay != -1:
if gen_cond_type == "batchnorm":
z_input = batchnorm(
[0], z_input,
is_training=is_training,
labels=input_clusters,
n_labels=clusters_no)
elif gen_cond_type == "layernorm":
z_input = layernorm([1],
z_input,
labels=input_clusters,
n_labels=clusters_no)
z_input = tf.nn.relu(z_input)
with tf.variable_scope("generator_layers_" + 'output'):
fake_outputs = layers.relu(
z_input, output_cells_dim,
weights_initializer=layers.variance_scaling_initializer(mode="FAN_AVG"),
biases_initializer=tf.zeros_initializer())
if output_lsn is not None:
gammas_output = tf.Variable(
np.ones(z_input.shape.as_list()[0]) * output_lsn,
trainable=False)
sigmas = tf.reduce_sum(fake_outputs, axis=1)
scale_ls = tf.cast(gammas_output, dtype=tf.float32) / \
(sigmas + sys.float_info.epsilon)
fake_outputs = tf.transpose(tf.transpose(fake_outputs) *
scale_ls)
return cls(fake_outputs, batch_size, latent_dim, output_cells_dim,
var_scope, gen_layers, output_lsn,
gen_cond_type=gen_cond_type, is_training=is_training,
clusters_ratios=clusters_ratios, clusters_no=clusters_no,
input_clusters=input_clusters, reuse=reuse)
def weight_variable(shape, stddev=0.05, name='weight'):
# initial=tf.truncated_normal(shape=shape,dtype=tf.float32,stddev=stddev)
initial = variance_scaling_initializer()
# return tf.Variable(initial,dtype=tf.float32,name=name)
return tf.Variable(initial(shape=shape), name=name)
def conv3d(input_, output_dim, k_h=3, k_w=3, k_d=3, d_h=2, d_w=2, d_d=3, padding='SAME', name="conv3d", with_w=False):
with tf.variable_scope(name):
w = tf.get_variable('w', [k_h, k_w, k_d, input_.get_shape()[-1], output_dim], initializer=variance_scaling_initializer())
if padding == 'Other':
# Not sure what's up with this r n --andrew
# Something about going from latent space to first conv.
padding = 'VALID'
input_ = tf.pad(input_, [[0,0], [3, 3], [3, 3], [3, 3], [0, 0]], "CONSTANT")
elif padding == 'VALID':
padding = 'VALID'
conv = tf.nn.conv3d(input_, w, strides=[1, d_h, d_w, d_d, 1], padding=padding)
biases = tf.get_variable('biases', [output_dim], initializer=tf.constant_initializer(0.0))
conv = tf.reshape(tf.nn.bias_add(conv, biases), conv.get_shape())
if with_w:
return conv, w, biases
else:
return conv
def conv2d(input, output_dim, ksize = [3, 3, 2, 2], padding='SAME', name="conv2d", with_w=False):
k_h, k_w, d_h, d_w =ksize[0], ksize[1], ksize[2], ksize[3]
with tf.variable_scope(name):
w = tf.get_variable('weights', [k_h, k_w, input.get_shape()[-1], output_dim], initializer= variance_scaling_initializer())
b = tf.get_variable('biases', [output_dim], initializer=tf.constant_initializer(0.0))
if padding == 'Other':
padding = 'VALID'
input = tf.pad(input, [[0,0], [3, 3], [3, 3], [0, 0]], "CONSTANT")
elif padding == 'VALID':
padding = 'VALID'
conv = tf.nn.conv2d(input, w, strides=[1, d_h, d_w, 1], padding=padding)
shape = conv.get_shape().as_list()
conv = tf.reshape(tf.nn.bias_add(conv, b), (-1,shape[1],shape[2],shape[3]))
if with_w:
return conv, w, b
else:
return conv
def fully_connect(input, output_size, stddev=0.02, scope=None, with_w=False):
with tf.variable_scope(scope or "Linear"):
w = tf.get_variable("weights", [input.get_shape().as_list()[1], output_size], tf.float32, variance_scaling_initializer())
b = tf.get_variable("biases", [output_size], initializer=tf.constant_initializer(0.0))
output = tf.matmul(input, w) + b
if with_w:
return output, w, b
else:
return output
def get_weights_initializer(self):
# return tf.truncated_normal_initializer(stddev=0.02)
return variance_scaling_initializer()
