def generator(input_var, Y):
yb = Y #.dimshuffle(0, 1, 'x', 'x')
G_1 = lasagne.layers.InputLayer(shape=(None, NLAT, 1, 1),
input_var=input_var)
G_2 = lasagne.layers.InputLayer(shape=(None, 10), input_var=Y)
network = G_1
network_yb = G_2
network = conv_layer(network, 1, 4 * 4 * 128, 1, 'valid')
#print(input_var.shape[0])
network = ll.ReshapeLayer(network, (-1, 128, 4, 4))
network = CondConvConcatLayer([network, network_yb])
network = resnet_block(network, 3, 138)
network = CondConvConcatLayer([network, network_yb])
#network = resnet_block(network, 3, 128)
network = BilinearUpsampling(network, ratio=2)
network = batch_norm(conv_layer(network, 3, 138, 1, 'same'))
network = CondConvConcatLayer([network, network_yb])
network = resnet_block(network, 3, 148)
network = CondConvConcatLayer([network, network_yb])
network = BilinearUpsampling(network, ratio=2)
network = batch_norm(conv_layer(network, 3, 32, 1, 'valid'))
network = BilinearUpsampling(network, ratio=2)
network = batch_norm(conv_layer(network, 3, 32, 1, 'same'))
network = CondConvConcatLayer([network, network_yb])
#network = resnet_block(network, 3, 32)
network = conv_layer(network, 1, 1, 1, 'valid', nonlinearity=sigmoid)
#network =lasagne.layers.Conv2DLayer(network, num_filters=1, filter_size=1, stride=1, nonlinearity=sigmoid)
return network, G_1, G_2
def generator(input_var):
network = lasagne.layers.InputLayer(shape=(None, NLAT, 1, 1),
input_var=input_var)
network = conv_layer(network, 1, 4 * 4 * 128, 1, 'valid')
#print(input_var.shape[0])
network = ll.ReshapeLayer(network, (-1, 128, 4, 4))
network = resnet_block(network, 3, 128)
network = resnet_block(network, 3, 128)
network = BilinearUpsampling(network, ratio=2)
network = batch_norm(conv_layer(network, 3, 64, 1, 'same'))
network = resnet_block(network, 3, 64)
network = BilinearUpsampling(network, ratio=2)
network = batch_norm(conv_layer(network, 3, 32, 1, 'valid'))
network = BilinearUpsampling(network, ratio=2)
network = resnet_block(network, 3, 32)
network = conv_layer(network, 1, 1, 1, 'valid', nonlinearity=sigmoid)
#network =lasagne.layers.Conv2DLayer(network, num_filters=1, filter_size=1, stride=1, nonlinearity=sigmoid)
return network
def discriminator(input_var, Y):
yb = Y.dimshuffle(0, 1, 'x', 'x')
D_1 = lasagne.layers.InputLayer(shape=(None, 1, 28, 28),
input_var=input_var)
D_2 = lasagne.layers.InputLayer(shape=(None, 10), input_var=Y)
network = D_1
network_yb = D_2
network = CondConvConcatLayer([network, network_yb])
network = ll.DropoutLayer(network, p=0.4)
network = conv_layer(network, 3, 32, 1, 'same', nonlinearity=lrelu)
network = CondConvConcatLayer([network, network_yb])
network = conv_layer(network, 3, 64, 2, 'same', nonlinearity=lrelu)
network = CondConvConcatLayer([network, network_yb])
network = conv_layer(network, 3, 64, 2, 'same', nonlinearity=lrelu)
#network = batch_norm(conv_layer(network, 3, 128, 1, 'same', nonlinearity=lrelu))
#network = ll.DropoutLayer(network, p=0.2)
network = conv_layer(network, 3, 128, 2, 'same', nonlinearity=lrelu)
network = CondConvConcatLayer([network, network_yb])
network = batch_norm(
conv_layer(network, 4, 128, 1, 'valid', nonlinearity=lrelu))
network = CondConvConcatLayer([network, network_yb])
#network= DropoutLayer(network, p=0.5)
network = conv_layer(network, 1, 1, 1, 'valid', nonlinearity=None)
return network, D_1, D_2
def discriminator(input_var):
network = lasagne.layers.InputLayer(shape=(None, 1, 28, 28),
input_var=input_var)
network = ll.DropoutLayer(network, p=0.5)
network = weight_norm(conv_layer(network,
3,
32,
1,
'same',
nonlinearity=lrelu),
train_g=False)
network = weight_norm(conv_layer(network,
3,
32,
2,
'same',
nonlinearity=lrelu),
train_g=False)
network = weight_norm(conv_layer(network,
3,
64,
2,
'same',
nonlinearity=lrelu),
train_g=False)
network = weight_norm(conv_layer(network,
3,
128,
2,
'same',
nonlinearity=lrelu),
train_g=False)
network = weight_norm(conv_layer(network,
4,
128,
1,
'valid',
nonlinearity=lrelu),
train_g=False)
network = weight_norm(conv_layer(network,
1,
1,
1,
'valid',
nonlinearity=None),
train_g=True)
return network