def test_backward_cpu_more_than_once(self):
func = functions.MaxPooling2D(3,
stride=2,
pad=1,
cover_all=self.cover_all)
func.apply(self.inputs)
func.backward((0, ), self.grad_outputs)
func.backward((0, ), self.grad_outputs)
def check_backward(self, x_data, y_grad, use_cudnn=True):
gradient_check.check_backward(
functions.MaxPooling2D(3,
stride=2,
pad=1,
cover_all=self.cover_all,
use_cudnn=use_cudnn), x_data, y_grad,
**self.check_backward_options)
def test_backward_cpu_more_than_once(self):
func = functions.MaxPooling2D(3,
stride=2,
pad=1,
cover_all=self.cover_all)
func.apply((self.x, ))
func.backward((0, ), (self.gy, ))
func.backward((0, ), (self.gy, ))
def check_backward(self, x_data, y_grad, use_cudnn='always'):
with chainer.using_config('use_cudnn', use_cudnn):
gradient_check.check_backward(
functions.MaxPooling2D(3,
stride=2,
pad=1,
cover_all=self.cover_all), x_data,
y_grad, **self.check_backward_options)
def __init__(self, in_channel, n_mid=64):
w = math.sqrt(2)
super(EncDec, self).__init__(
enc=L.Convolution2D(in_channel, n_mid, 7, 1, 3, w),
bn_m=L.BatchNormalization(n_mid),
dec=L.Convolution2D(n_mid, n_mid, 7, 1, 3, w),
bn_o=L.BatchNormalization(n_mid),
)
self.p = F.MaxPooling2D(2, 2, use_cudnn=False)
self.inside = None
def __init__(self, n_channels, n_filters=8, ksize=5, task_is_train=True):
self.dtype = np.float32
W = chainer.initializers.HeNormal(1 / np.sqrt(2), self.dtype)
bias = chainer.initializers.Zero(self.dtype)
self.pooling_ksize = 2
self.pooling_stride = 2
self.h1_size = None
self.h2_size = None
self.h3_size = None
super(ConvolutionalVariationalAutoencoder, self).__init__(
# encoder
enc_conv1=L.Convolution2D(n_channels, out_channels=n_filters, ksize=ksize, pad=ksize // 2,
initialW=W, initial_bias=bias),
enc_bn1=L.BatchNormalization(n_filters),
enc_conv2=L.Convolution2D(n_filters, n_filters // 2, ksize=ksize, pad=ksize // 2,
initialW=W, initial_bias=bias),
enc_bn2=L.BatchNormalization(n_filters // 2),
enc_conv3=L.Convolution2D(n_filters // 2, n_filters // 2, ksize=ksize, pad=ksize // 2,
initialW=W, initial_bias=bias),
enc_bn3=L.BatchNormalization(n_filters // 2),
enc_conv4_mu=L.Convolution2D(n_filters // 2, n_filters // 2, ksize=ksize, pad=ksize // 2,
initialW=W, initial_bias=bias),
enc_conv4_ln_var=L.Convolution2D(n_filters // 2, n_filters // 2, ksize=ksize, pad=ksize // 2,
initialW=W, initial_bias=bias),
# decoder
dec_conv1=L.Convolution2D(n_filters // 2, n_filters // 2, ksize=ksize, pad=ksize // 2,
initialW=W, initial_bias=bias),
dec_bn1=L.BatchNormalization(n_filters // 2),
dec_conv2=L.Convolution2D(n_filters // 2, n_filters // 2, ksize=ksize, pad=ksize // 2,
initialW=W, initial_bias=bias),
dec_bn2=L.BatchNormalization(n_filters // 2),
dec_conv3=L.Convolution2D(n_filters // 2, n_filters, ksize=ksize, pad=ksize // 2,
initialW=W, initial_bias=bias),
dec_bn3=L.BatchNormalization(n_filters),
dec_conv4_mu=L.Convolution2D(n_filters, n_channels, ksize=ksize, pad=ksize // 2,
initialW=W, initial_bias=bias),
dec_conv4_ln_var=L.Convolution2D(n_filters, n_channels, ksize=ksize, pad=ksize // 2,
initialW=W, initial_bias=bias)
)
self.p1 = F.MaxPooling2D(2, 2)
self.p2 = F.MaxPooling2D(2, 2)
self.p3 = F.MaxPooling2D(2, 2)
self.train = task_is_train
def __init__(self, train, alphabet_size, feature_size=256):
self.feature_size = feature_size
super(IngredientSentenceSixLayers, self).__init__(
train=train,
alphabet_size=alphabet_size,
out_feature=self.feature_size * 6,
conv0=L.Convolution2D(alphabet_size,
self.feature_size, (1, 7),
stride=1,
pad=0),
conv1=L.Convolution2D(self.feature_size,
self.feature_size, (1, 5),
stride=1,
pad=0),
conv2=L.Convolution2D(self.feature_size,
self.feature_size, (1, 1),
stride=1,
pad=0),
conv3=L.Convolution2D(self.feature_size,
self.feature_size, (1, 3),
stride=1,
pad=0),
conv4=L.Convolution2D(self.feature_size,
self.feature_size, (1, 1),
stride=1,
pad=0),
conv5=L.Convolution2D(self.feature_size,
self.feature_size, (1, 1),
stride=1,
pad=0),
)
# We want to use 1d max-pooling, such that ingredient information is not crossed and stays in tact.
# The default Functions.max_pooling_nd() function will
# always take the full input(-2) dimensions of the x input.
# In our case this would mean that it would be 2d. Therefore we have to use the class manually.
# Since our dimensionality be 4 : BATCH_SIZE X RECIPE_SIZE X TEXT_SIZE X FEATURE_SIZE
self.maxp3 = F.MaxPooling2D(
(1, 3)) # We will only use max pooling with kernel size 3
self.maxp2 = F.MaxPooling2D(
(1, 2)) # We will only use max pooling with kernel size 3
def setUp(self):
self.x = numpy.random.uniform(-1, 1, self.in_shape).astype(self.dtype)
self.p = F.MaxPooling2D(2, 2)
with chainer.using_config('use_cudnn', 'never'):
self.pooled_y = self.p.apply((self.x, ))[0]
self.gy = numpy.random.uniform(-1, 1, self.in_shape).astype(self.dtype)
self.ggx = numpy.random.uniform(-1, 1,
self.pooled_y.shape).astype(self.dtype)
self.check_backward_options = {}
self.check_double_backward_options = {'atol': 1e-3, 'rtol': 1e-2}
if self.dtype == numpy.float16:
self.check_double_backward_options = {'atol': 3e-3, 'rtol': 3e-2}
def __init__(self,
in_ch=3,
mid_ch=0,
out_ch=13,
ksize=3,
stride=1,
pad=1,
dilation=1,
nobias=False,
symmetric=False,
drop_ratio=0.0,
downsample=False,
upsample=False,
p=None,
use_bn=True,
train=True):
super(Block, self).__init__()
k1, k2, s1 = self.calc_param(downsample, symmetric, upsample)
self.p = p
self.drop_ratio = drop_ratio
self.downsample = downsample
self.upsample = upsample
self.train = train
with self.init_scope():
this_mod = sys.modules[__name__]
conv_type = "ConvBN" if use_bn else "Conv"
ConvBlock = getattr(this_mod, conv_type + "PReLU")
self.conv1 = ConvBlock(in_ch, mid_ch, k1, s1, 0, nobias=True)
conv_type2 = conv_type + "PReLU"
conv_type2 = "Symmetric" + conv_type2 if symmetric else conv_type2
ConvBlock = getattr(this_mod, conv_type2)
self.conv2 = ConvBlock(mid_ch,
mid_ch,
k2,
stride,
pad,
dilation,
nobias=False,
upsample=upsample)
ConvBlock = getattr(this_mod, conv_type)
self.conv3 = ConvBlock(mid_ch, out_ch, 1, 1, 0, nobias=True)
self.prelu = L.PReLU()
if downsample:
ConvBlock = getattr(this_mod, conv_type)
self.conv = ConvBlock(in_ch, out_ch, 1, 1, 0, nobias=True)
self.p = F.MaxPooling2D(2, 2)
if upsample:
ConvBlock = getattr(this_mod, conv_type)
self.conv = ConvBlock(in_ch, out_ch, 1, 1, 0, nobias=True)
def __init__(self, train=False):
super(VGG16, self).__init__()
self.trunk = [
('conv1_1', L.Convolution2D(3, 64, 3, 1, 1)),
('relu1_1', F.ReLU()),
('conv1_2', L.Convolution2D(64, 64, 3, 1, 1)),
('relu1_2', F.ReLU()),
('pool1', F.MaxPooling2D(2, 2)),
('conv2_1', L.Convolution2D(64, 128, 3, 1, 1)),
('relu2_1', F.ReLU()),
('conv2_2', L.Convolution2D(128, 128, 3, 1, 1)),
('relu2_2', F.ReLU()),
('pool2', F.MaxPooling2D(2, 2)),
('conv3_1', L.Convolution2D(128, 256, 3, 1, 1)),
('relu3_1', F.ReLU()),
('conv3_2', L.Convolution2D(256, 256, 3, 1, 1)),
('relu3_2', F.ReLU()),
('conv3_3', L.Convolution2D(256, 256, 3, 1, 1)),
('relu3_3', F.ReLU()),
('pool3', F.MaxPooling2D(2, 2)),
('conv4_1', L.Convolution2D(256, 512, 3, 1, 1)),
('relu4_1', F.ReLU()),
('conv4_2', L.Convolution2D(512, 512, 3, 1, 1)),
('relu4_2', F.ReLU()),
('conv4_3', L.Convolution2D(512, 512, 3, 1, 1)),
('relu4_3', F.ReLU()),
('pool4', F.MaxPooling2D(2, 2)),
('conv5_1', L.Convolution2D(512, 512, 3, 1, 1)),
('relu5_1', F.ReLU()),
('conv5_2', L.Convolution2D(512, 512, 3, 1, 1)),
('relu5_2', F.ReLU()),
('conv5_3', L.Convolution2D(512, 512, 3, 1, 1)),
('relu5_3', F.ReLU()),
('rpn_conv_3x3', L.Convolution2D(512, 512, 3, 1, 1)),
('rpn_relu_3x3', F.ReLU()),
]
for name, link in self.trunk:
if 'conv' in name:
self.add_link(name, link)
def __init__(self, train=False):
super(VGG16Prev, self).__init__()
self.trunk = [
('conv1_1', L.Convolution2D(3, 64, 3, 1, 1)),
('_relu1_1', F.ReLU()),
('conv1_2', L.Convolution2D(64, 64, 3, 1, 1)),
('_relu1_2', F.ReLU()),
('_pool1', F.MaxPooling2D(2, 2)),
('conv2_1', L.Convolution2D(64, 128, 3, 1, 1)),
('_relu2_1', F.ReLU()),
('conv2_2', L.Convolution2D(128, 128, 3, 1, 1)),
('_relu2_2', F.ReLU()),
('_pool2', F.MaxPooling2D(2, 2)),
('conv3_1', L.Convolution2D(128, 256, 3, 1, 1)),
('_relu3_1', F.ReLU()),
('conv3_2', L.Convolution2D(256, 256, 3, 1, 1)),
('_relu3_2', F.ReLU()),
('conv3_3', L.Convolution2D(256, 256, 3, 1, 1)),
('_relu3_3', F.ReLU()),
('_pool3', F.MaxPooling2D(2, 2)),
('conv4_1', L.Convolution2D(256, 512, 3, 1, 1)),
('_relu4_1', F.ReLU()),
('conv4_2', L.Convolution2D(512, 512, 3, 1, 1)),
('_relu4_2', F.ReLU()),
('conv4_3', L.Convolution2D(512, 512, 3, 1, 1)),
('_relu4_3', F.ReLU()),
('_pool4', F.MaxPooling2D(2, 2)),
('conv5_1', L.Convolution2D(512, 512, 3, 1, 1)),
('_relu5_1', F.ReLU()),
('conv5_2', L.Convolution2D(512, 512, 3, 1, 1)),
('_relu5_2', F.ReLU()),
('conv5_3', L.Convolution2D(512, 512, 3, 1, 1)),
('_relu5_3', F.ReLU()),
]
for name, link in self.trunk:
if not name.startswith('_'):
self.add_link(name, link)
def __init__(self, arch, dr=0.5, bn=False):
super(discriminator, self).__init__()
self.dr = dr
self.train = True
self.best_score = collections.defaultdict(lambda: 0)
self.functions = collections.OrderedDict()
for l in arch['0']:
name = l['name']
if reg_conv.match(name):
self.add_link(
name,
L.Convolution2D(l['in_channel'],
l['out_channel'],
l['ksize'],
pad=l['pad']))
if bn:
m = reg_num.search(name)
bn_name = 'cbn_layer{}'.format(m.group())
self.add_link(bn_name,
L.BatchNormalization(l['out_channel']))
self.functions[name] = [self[name], self[bn_name], F.relu]
else:
self.functions[name] = [self[name], F.relu]
if reg_pool.match(name):
self.functions[name] = [
F.MaxPooling2D(l['ksize'], l['stride'])
]
if reg_fc.match(name):
self.add_link(name, L.Linear(l['in_size'], l['out_size']))
if bn:
m = reg_num.search(name)
bn_name = 'fbn_layer{}'.format(m.group())
self.add_link(bn_name, L.BatchNormalization(l['out_size']))
self.functions[name] = [
self[name], self[bn_name], F.relu, F.dropout
]
else:
self.functions[name] = [self[name], F.relu, F.dropout]
if name == 'out_layer':
self.add_link(name, L.Linear(l['in_size'], l['out_size']))
self.functions[name] = [self[name]]
self.functions['prob'] = [F.softmax]
def check_backward_consistency_regression(self,
x_data,
gy_data,
use_cudnn=True):
# Regression test to two-dimensional max pooling layer.
if len(self.dims) != 2:
return
ksize = self.ksize
stride = self.stride
pad = self.pad
xp = cuda.get_array_module(x_data)
# Backward computation for N-dimensional max pooling layer.
x_nd = chainer.Variable(xp.array(x_data))
func_nd = functions.MaxPoolingND(self.ndim,
ksize,
stride=stride,
pad=pad,
use_cudnn=use_cudnn,
cover_all=self.cover_all)
y_nd = func_nd(x_nd)
y_nd.grad = gy_data
y_nd.backward()
# Backward computation for two-dimensional max pooling layer.
x_2d = chainer.Variable(xp.array(x_data))
func_2d = functions.MaxPooling2D(ksize,
stride=stride,
pad=pad,
use_cudnn=use_cudnn,
cover_all=self.cover_all)
y_2d = func_2d(x_2d)
y_2d.grad = gy_data
y_2d.backward()
# Test that the two result gradients are close enough.
testing.assert_allclose(x_nd.grad, x_2d.grad)
def __init__(self):
super().__init__(
conv=L.Convolution2D(3, 32, 3, stride=2, pad=0),
conv_1=L.Convolution2D(32, 32, 3, stride=1, pad=0),
conv_2=L.Convolution2D(32, 64, 3, stride=1, pad=1),
conv_3=L.Convolution2D(64, 80, 1, stride=1, pad=0),
conv_4=L.Convolution2D(80, 192, 3, stride=1, pad=0),
bn_conv=L.BatchNormalization(32),
bn_conv_1=L.BatchNormalization(32),
bn_conv_2=L.BatchNormalization(64),
bn_conv_3=L.BatchNormalization(80),
bn_conv_4=L.BatchNormalization(192),
mixed=Mixed([
('conv',
Tower([('conv', L.Convolution2D(192, 64, 1, stride=1, pad=0)),
('bn_conv', L.BatchNormalization(64)),
('_relu', F.ReLU())])),
('tower',
Tower([('conv', L.Convolution2D(192, 48, 1, stride=1, pad=0)),
('bn_conv', L.BatchNormalization(48)),
('_relu', F.ReLU()),
('conv_1', L.Convolution2D(48, 64, 5, stride=1,
pad=2)),
('bn_conv_1', L.BatchNormalization(64)),
('_relu_1', F.ReLU())])),
('tower_1',
Tower([('conv', L.Convolution2D(192, 64, 1, stride=1, pad=0)),
('bn_conv', L.BatchNormalization(64)),
('_relu', F.ReLU()),
('conv_1', L.Convolution2D(64, 96, 3, stride=1,
pad=1)),
('bn_conv_1', L.BatchNormalization(96)),
('_relu_1', F.ReLU()),
('conv_2', L.Convolution2D(96, 96, 3, stride=1,
pad=1)),
('bn_conv_2', L.BatchNormalization(96)),
('_relu_2', F.ReLU())])),
('tower_2',
Tower([('_pooling', F.AveragePooling2D(3, 1, pad=1)),
('conv', L.Convolution2D(192, 32, 1, stride=1, pad=0)),
('bn_conv', L.BatchNormalization(32)),
('_relu', F.ReLU())]))
]),
mixed_1=Mixed([
('conv',
Tower([('conv', L.Convolution2D(256, 64, 1, stride=1, pad=0)),
('bn_conv', L.BatchNormalization(64)),
('_relu', F.ReLU())])),
('tower',
Tower([('conv', L.Convolution2D(256, 48, 1, stride=1, pad=0)),
('bn_conv', L.BatchNormalization(48)),
('_relu', F.ReLU()),
('conv_1', L.Convolution2D(48, 64, 5, stride=1,
pad=2)),
('bn_conv_1', L.BatchNormalization(64)),
('_relu_1', F.ReLU())])),
('tower_1',
Tower([('conv', L.Convolution2D(256, 64, 1, stride=1, pad=0)),
('bn_conv', L.BatchNormalization(64)),
('_relu', F.ReLU()),
('conv_1', L.Convolution2D(64, 96, 3, stride=1,
pad=1)),
('bn_conv_1', L.BatchNormalization(96)),
('_relu_1', F.ReLU()),
('conv_2', L.Convolution2D(96, 96, 3, stride=1,
pad=1)),
('bn_conv_2', L.BatchNormalization(96)),
('_relu_2', F.ReLU())])),
('tower_2',
Tower([('_pooling', F.AveragePooling2D(3, 1, pad=1)),
('conv', L.Convolution2D(256, 64, 1, stride=1, pad=0)),
('bn_conv', L.BatchNormalization(64)),
('_relu', F.ReLU())]))
]),
mixed_2=Mixed([
('conv',
Tower([('conv', L.Convolution2D(288, 64, 1, stride=1, pad=0)),
('bn_conv', L.BatchNormalization(64)),
('_relu', F.ReLU())])),
('tower',
Tower([('conv', L.Convolution2D(288, 48, 1, stride=1, pad=0)),
('bn_conv', L.BatchNormalization(48)),
('_relu', F.ReLU()),
('conv_1', L.Convolution2D(48, 64, 5, stride=1,
pad=2)),
('bn_conv_1', L.BatchNormalization(64)),
('_relu_1', F.ReLU())])),
('tower_1',
Tower([('conv', L.Convolution2D(288, 64, 1, stride=1, pad=0)),
('bn_conv', L.BatchNormalization(64)),
('_relu', F.ReLU()),
('conv_1', L.Convolution2D(64, 96, 3, stride=1,
pad=1)),
('bn_conv_1', L.BatchNormalization(96)),
('_relu_1', F.ReLU()),
('conv_2', L.Convolution2D(96, 96, 3, stride=1,
pad=1)),
('bn_conv_2', L.BatchNormalization(96)),
('_relu_2', F.ReLU())])),
('tower_2',
Tower([('_pooling', F.AveragePooling2D(3, 1, pad=1)),
('conv', L.Convolution2D(288, 64, 1, stride=1, pad=0)),
('bn_conv', L.BatchNormalization(64)),
('_relu', F.ReLU())]))
]),
mixed_3=Mixed([
('conv',
Tower([('conv', L.Convolution2D(288, 384, 3, stride=2,
pad=0)),
('bn_conv', L.BatchNormalization(384)),
('_relu', F.ReLU())])),
('tower',
Tower([('conv', L.Convolution2D(288, 64, 1, stride=1, pad=0)),
('bn_conv', L.BatchNormalization(64)),
('_relu', F.ReLU()),
('conv_1', L.Convolution2D(64, 96, 3, stride=1,
pad=1)),
('bn_conv_1', L.BatchNormalization(96)),
('_relu_1', F.ReLU()),
('conv_2', L.Convolution2D(96, 96, 3, stride=2,
pad=0)),
('bn_conv_2', L.BatchNormalization(96)),
('_relu_2', F.ReLU())])),
('pool', Tower([('_pooling', F.MaxPooling2D(3, 2, pad=0))]))
]),
mixed_4=Mixed([
('conv',
Tower([('conv', L.Convolution2D(768, 192, 1, stride=1,
pad=0)),
('bn_conv', L.BatchNormalization(192)),
('_relu', F.ReLU())])),
('tower',
Tower([('conv', L.Convolution2D(768, 128, 1, stride=1,
pad=0)),
('bn_conv', L.BatchNormalization(128)),
('_relu', F.ReLU()),
('conv_1',
L.Convolution2D(128,
128, (1, 7),
stride=1,
pad=(0, 3))),
('bn_conv_1', L.BatchNormalization(128)),
('_relu_1', F.ReLU()),
('conv_2',
L.Convolution2D(128,
192, (7, 1),
stride=1,
pad=(3, 0))),
('bn_conv_2', L.BatchNormalization(192)),
('_relu_2', F.ReLU())])),
('tower_1',
Tower([('conv', L.Convolution2D(768, 128, 1, stride=1,
pad=0)),
('bn_conv', L.BatchNormalization(128)),
('_relu', F.ReLU()),
('conv_1',
L.Convolution2D(128,
128, (7, 1),
stride=1,
pad=(3, 0))),
('bn_conv_1', L.BatchNormalization(128)),
('_relu_1', F.ReLU()),
('conv_2',
L.Convolution2D(128,
128, (1, 7),
stride=1,
pad=(0, 3))),
('bn_conv_2', L.BatchNormalization(128)),
('_relu_2', F.ReLU()),
('conv_3',
L.Convolution2D(128,
128, (7, 1),
stride=1,
pad=(3, 0))),
('bn_conv_3', L.BatchNormalization(128)),
('_relu_3', F.ReLU()),
('conv_4',
L.Convolution2D(128,
192, (1, 7),
stride=1,
pad=(0, 3))),
('bn_conv_4', L.BatchNormalization(192)),
('_relu_4', F.ReLU())])),
('tower_2',
Tower([('_pooling', F.AveragePooling2D(3, 1, pad=1)),
('conv', L.Convolution2D(768, 192, 1, stride=1,
pad=0)),
('bn_conv', L.BatchNormalization(192)),
('_relu', F.ReLU())]))
]),
mixed_5=Mixed([
('conv',
Tower([('conv', L.Convolution2D(768, 192, 1, stride=1,
pad=0)),
('bn_conv', L.BatchNormalization(192)),
('_relu', F.ReLU())])),
('tower',
Tower([('conv', L.Convolution2D(768, 160, 1, stride=1,
pad=0)),
('bn_conv', L.BatchNormalization(160)),
('_relu', F.ReLU()),
('conv_1',
L.Convolution2D(160,
160, (1, 7),
stride=1,
pad=(0, 3))),
('bn_conv_1', L.BatchNormalization(160)),
('_relu_1', F.ReLU()),
('conv_2',
L.Convolution2D(160,
192, (7, 1),
stride=1,
pad=(3, 0))),
('bn_conv_2', L.BatchNormalization(192)),
('_relu_2', F.ReLU())])),
('tower_1',
Tower([('conv', L.Convolution2D(768, 160, 1, stride=1,
pad=0)),
('bn_conv', L.BatchNormalization(160)),
('_relu', F.ReLU()),
('conv_1',
L.Convolution2D(160,
160, (7, 1),
stride=1,
pad=(3, 0))),
('bn_conv_1', L.BatchNormalization(160)),
('_relu_1', F.ReLU()),
('conv_2',
L.Convolution2D(160,
160, (1, 7),
stride=1,
pad=(0, 3))),
('bn_conv_2', L.BatchNormalization(160)),
('_relu_2', F.ReLU()),
('conv_3',
L.Convolution2D(160,
160, (7, 1),
stride=1,
pad=(3, 0))),
('bn_conv_3', L.BatchNormalization(160)),
('_relu_3', F.ReLU()),
('conv_4',
L.Convolution2D(160,
192, (1, 7),
stride=1,
pad=(0, 3))),
('bn_conv_4', L.BatchNormalization(192)),
('_relu_4', F.ReLU())])),
('tower_2',
Tower([('_pooling', F.AveragePooling2D(3, 1, pad=1)),
('conv', L.Convolution2D(768, 192, 1, stride=1,
pad=0)),
('bn_conv', L.BatchNormalization(192)),
('_relu', F.ReLU())]))
]),
mixed_6=Mixed([
('conv',
Tower([('conv', L.Convolution2D(768, 192, 1, stride=1,
pad=0)),
('bn_conv', L.BatchNormalization(192)),
('_relu', F.ReLU())])),
('tower',
Tower([('conv', L.Convolution2D(768, 160, 1, stride=1,
pad=0)),
('bn_conv', L.BatchNormalization(160)),
('_relu', F.ReLU()),
('conv_1',
L.Convolution2D(160,
160, (1, 7),
stride=1,
pad=(0, 3))),
('bn_conv_1', L.BatchNormalization(160)),
('_relu_1', F.ReLU()),
('conv_2',
L.Convolution2D(160,
192, (7, 1),
stride=1,
pad=(3, 0))),
('bn_conv_2', L.BatchNormalization(192)),
('_relu_2', F.ReLU())])),
('tower_1',
Tower([('conv', L.Convolution2D(768, 160, 1, stride=1,
pad=0)),
('bn_conv', L.BatchNormalization(160)),
('_relu', F.ReLU()),
('conv_1',
L.Convolution2D(160,
160, (7, 1),
stride=1,
pad=(3, 0))),
('bn_conv_1', L.BatchNormalization(160)),
('_relu_1', F.ReLU()),
('conv_2',
L.Convolution2D(160,
160, (1, 7),
stride=1,
pad=(0, 3))),
('bn_conv_2', L.BatchNormalization(160)),
('_relu_2', F.ReLU()),
('conv_3',
L.Convolution2D(160,
160, (7, 1),
stride=1,
pad=(3, 0))),
('bn_conv_3', L.BatchNormalization(160)),
('_relu_3', F.ReLU()),
('conv_4',
L.Convolution2D(160,
192, (1, 7),
stride=1,
pad=(0, 3))),
('bn_conv_4', L.BatchNormalization(192)),
('_relu_4', F.ReLU())])),
('tower_2',
Tower([('_pooling', F.AveragePooling2D(3, 1, pad=1)),
('conv', L.Convolution2D(768, 192, 1, stride=1,
pad=0)),
('bn_conv', L.BatchNormalization(192)),
('_relu', F.ReLU())]))
]),
mixed_7=Mixed([
('conv',
Tower([('conv', L.Convolution2D(768, 192, 1, stride=1,
pad=0)),
('bn_conv', L.BatchNormalization(192)),
('_relu', F.ReLU())])),
('tower',
Tower([('conv', L.Convolution2D(768, 192, 1, stride=1,
pad=0)),
('bn_conv', L.BatchNormalization(192)),
('_relu', F.ReLU()),
('conv_1',
L.Convolution2D(192,
192, (1, 7),
stride=1,
pad=(0, 3))),
('bn_conv_1', L.BatchNormalization(192)),
('_relu_1', F.ReLU()),
('conv_2',
L.Convolution2D(192,
192, (7, 1),
stride=1,
pad=(3, 0))),
('bn_conv_2', L.BatchNormalization(192)),
('_relu_2', F.ReLU())])),
('tower_1',
Tower([('conv', L.Convolution2D(768, 192, 1, stride=1,
pad=0)),
('bn_conv', L.BatchNormalization(192)),
('_relu', F.ReLU()),
('conv_1',
L.Convolution2D(192,
192, (7, 1),
stride=1,
pad=(3, 0))),
('bn_conv_1', L.BatchNormalization(192)),
('_relu_1', F.ReLU()),
('conv_2',
L.Convolution2D(192,
192, (1, 7),
stride=1,
pad=(0, 3))),
('bn_conv_2', L.BatchNormalization(192)),
('_relu_2', F.ReLU()),
('conv_3',
L.Convolution2D(192,
192, (7, 1),
stride=1,
pad=(3, 0))),
('bn_conv_3', L.BatchNormalization(192)),
('_relu_3', F.ReLU()),
('conv_4',
L.Convolution2D(192,
192, (1, 7),
stride=1,
pad=(0, 3))),
('bn_conv_4', L.BatchNormalization(192)),
('_relu_4', F.ReLU())])),
('tower_2',
Tower([('_pooling', F.AveragePooling2D(3, 1, pad=1)),
('conv', L.Convolution2D(768, 192, 1, stride=1,
pad=0)),
('bn_conv', L.BatchNormalization(192)),
('_relu', F.ReLU())]))
]),
mixed_8=Mixed([
('tower',
Tower([('conv', L.Convolution2D(768, 192, 1, stride=1,
pad=0)),
('bn_conv', L.BatchNormalization(192)),
('_relu', F.ReLU()),
('conv_1', L.Convolution2D(192,
320,
3,
stride=2,
pad=0)),
('bn_conv_1', L.BatchNormalization(320)),
('_relu_1', F.ReLU())])),
('tower_1',
Tower([('conv', L.Convolution2D(768, 192, 1, stride=1,
pad=0)),
('bn_conv', L.BatchNormalization(192)),
('_relu', F.ReLU()),
('conv_1',
L.Convolution2D(192,
192, (1, 7),
stride=1,
pad=(0, 3))),
('bn_conv_1', L.BatchNormalization(192)),
('_relu_1', F.ReLU()),
('conv_2',
L.Convolution2D(192,
192, (7, 1),
stride=1,
pad=(3, 0))),
('bn_conv_2', L.BatchNormalization(192)),
('_relu_2', F.ReLU()),
('conv_3', L.Convolution2D(192,
192,
3,
stride=2,
pad=0)),
('bn_conv_3', L.BatchNormalization(192)),
('_relu_3', F.ReLU())])),
('pool', Tower([('_pooling', F.MaxPooling2D(3, 2, pad=0))]))
]),
mixed_9=Mixed([
('conv',
Tower([
('conv', L.Convolution2D(1280, 320, 1, stride=1, pad=0)),
('bn_conv', L.BatchNormalization(320)),
('_relu', F.ReLU()),
])),
('tower',
Tower([('conv', L.Convolution2D(1280, 384, 1, stride=1,
pad=0)),
('bn_conv', L.BatchNormalization(384)),
('_relu', F.ReLU()),
('mixed',
Mixed([('conv',
Tower([
('conv',
L.Convolution2D(384,
384, (1, 3),
stride=1,
pad=(0, 1))),
('bn_conv', L.BatchNormalization(384)),
('_relu', F.ReLU()),
])),
('conv_1',
Tower([
('conv_1',
L.Convolution2D(384,
384, (3, 1),
stride=1,
pad=(1, 0))),
('bn_conv_1', L.BatchNormalization(384)),
('_relu_1', F.ReLU()),
]))]))])),
('tower_1',
Tower([('conv', L.Convolution2D(1280, 448, 1, stride=1,
pad=0)),
('bn_conv', L.BatchNormalization(448)),
('_relu', F.ReLU()),
('conv_1', L.Convolution2D(448,
384,
3,
stride=1,
pad=1)),
('bn_conv_1', L.BatchNormalization(384)),
('_relu_1', F.ReLU()),
('mixed',
Mixed([('conv',
Tower([
('conv',
L.Convolution2D(384,
384, (1, 3),
stride=1,
pad=(0, 1))),
('bn_conv', L.BatchNormalization(384)),
('_relu', F.ReLU()),
])),
('conv_1',
Tower([
('conv_1',
L.Convolution2D(384,
384, (3, 1),
stride=1,
pad=(1, 0))),
('bn_conv_1', L.BatchNormalization(384)),
('_relu_1', F.ReLU()),
]))]))])),
('tower_2',
Tower([('_pooling', F.AveragePooling2D(3, 1, pad=1)),
('conv', L.Convolution2D(1280, 192, 1, stride=1,
pad=0)),
('bn_conv', L.BatchNormalization(192)),
('_relu', F.ReLU())]))
]),
mixed_10=Mixed([
('conv',
Tower([
('conv', L.Convolution2D(2048, 320, 1, stride=1, pad=0)),
('bn_conv', L.BatchNormalization(320)),
('_relu', F.ReLU()),
])),
('tower',
Tower([('conv', L.Convolution2D(2048, 384, 1, stride=1,
pad=0)),
('bn_conv', L.BatchNormalization(384)),
('_relu', F.ReLU()),
('mixed',
Mixed([('conv',
Tower([
('conv',
L.Convolution2D(384,
384, (1, 3),
stride=1,
pad=(0, 1))),
('bn_conv', L.BatchNormalization(384)),
('_relu', F.ReLU()),
])),
('conv_1',
Tower([
('conv_1',
L.Convolution2D(384,
384, (3, 1),
stride=1,
pad=(1, 0))),
('bn_conv_1', L.BatchNormalization(384)),
('_relu_1', F.ReLU()),
]))]))])),
('tower_1',
Tower([('conv', L.Convolution2D(2048, 448, 1, stride=1,
pad=0)),
('bn_conv', L.BatchNormalization(448)),
('_relu', F.ReLU()),
('conv_1', L.Convolution2D(448,
384,
3,
stride=1,
pad=1)),
('bn_conv_1', L.BatchNormalization(384)),
('_relu_1', F.ReLU()),
('mixed',
Mixed([('conv',
Tower([('conv',
L.Convolution2D(384,
384, (1, 3),
stride=1,
pad=(0, 1))),
('bn_conv', L.BatchNormalization(384)),
('_relu', F.ReLU())])),
('conv_1',
Tower([('conv_1',
L.Convolution2D(384,
384, (3, 1),
stride=1,
pad=(1, 0))),
('bn_conv_1',
L.BatchNormalization(384)),
('_relu_1', F.ReLU())]))]))])),
('tower_2',
Tower([('_pooling', F.MaxPooling2D(3, 1, pad=1)),
('conv', L.Convolution2D(2048, 192, 1, stride=1,
pad=0)),
('bn_conv', L.BatchNormalization(192)),
('_relu', F.ReLU())]))
]),
logit=L.Linear(2048, 1008))
def setUp(self):
self.x = numpy.random.uniform(-1, 1, self.in_shape).astype('f')
self.p = F.MaxPooling2D(2, 2, use_cudnn=False)
self.pooled_y = self.p(self.x)
self.gy = numpy.random.uniform(
-1, 1, self.in_shape).astype(numpy.float32)
def setUp(self):
self.x = numpy.random.uniform(-1, 1, self.in_shape).astype(self.dtype)
self.p = F.MaxPooling2D(2, 2)
with chainer.using_config('use_cudnn', 'never'):
self.pooled_y = self.p(self.x)
self.gy = numpy.random.uniform(-1, 1, self.in_shape).astype(self.dtype)
import chainer.functions as F
import numpy as np
import time
# input = [ [ 1, 2, 3 ],
# [ 4, 5, 6 ],
# [ 7, 8, 9 ] ]
repeat = 1
(n, c, h, w) = (64, 256, 256, 256)
x = np.ones((n, c, h, w), dtype=np.float32)
f = F.MaxPooling2D(3, stride=2, pad=0, use_cudnn=False)
x = x,
y = f.forward_cpu(x)
start = time.time()
for i in range(repeat):
y = f.forward_cpu(x)
print("average forward %f seconds" % ((time.time() - start) / repeat))
gy = np.ndarray(y[0].shape, dtype=np.float32)
gy.fill(0.001)
gy = gy,
gx = f.backward_cpu(x, gy)
start = time.time()
for i in range(repeat):
gx = f.backward_cpu(x, gy)
print("average backward %f seconds" % ((time.time() - start) / repeat))
def __init__(self,
cgp,
n_class,
lossfun=softmax_cross_entropy.softmax_cross_entropy,
accfun=accuracy.accuracy):
super(CGP2CNN, self).__init__()
self.cgp = cgp
self.pool_size = 2
initializer = chainer.initializers.HeNormal()
links = []
i = 1
for name, in1, in2 in self.cgp:
if name == 'pool_max':
links += [('_' + name + '_' + str(i),
F.MaxPooling2D(self.pool_size, self.pool_size, 0,
False))]
elif name == 'pool_ave':
links += [('_' + name + '_' + str(i),
F.AveragePooling2D(self.pool_size, self.pool_size,
0, False))]
elif name == 'concat':
links += [('_' + name + '_' + str(i), F.Concat())]
elif name == 'sum':
links += [('_' + name + '_' + str(i), F.Concat())
] # the F.Concat() is dummy
elif name == 'ConvBlock32_3':
links += [(name + '_' + str(i), ConvBlock(3, 32, initializer))]
elif name == 'ConvBlock32_5':
links += [(name + '_' + str(i), ConvBlock(5, 32, initializer))]
elif name == 'ConvBlock32_7':
links += [(name + '_' + str(i), ConvBlock(7, 32, initializer))]
elif name == 'ConvBlock64_3':
links += [(name + '_' + str(i), ConvBlock(3, 64, initializer))]
elif name == 'ConvBlock64_5':
links += [(name + '_' + str(i), ConvBlock(5, 64, initializer))]
elif name == 'ConvBlock64_7':
links += [(name + '_' + str(i), ConvBlock(7, 64, initializer))]
elif name == 'ConvBlock128_3':
links += [(name + '_' + str(i), ConvBlock(3, 128,
initializer))]
elif name == 'ConvBlock128_5':
links += [(name + '_' + str(i), ConvBlock(5, 128,
initializer))]
elif name == 'ConvBlock128_7':
links += [(name + '_' + str(i), ConvBlock(7, 128,
initializer))]
elif name == 'ResBlock32_3':
links += [(name + '_' + str(i), ResBlock(3, 32, initializer))]
elif name == 'ResBlock32_5':
links += [(name + '_' + str(i), ResBlock(5, 32, initializer))]
elif name == 'ResBlock32_7':
links += [(name + '_' + str(i), ResBlock(7, 32, initializer))]
elif name == 'ResBlock64_3':
links += [(name + '_' + str(i), ResBlock(3, 64, initializer))]
elif name == 'ResBlock64_5':
links += [(name + '_' + str(i), ResBlock(5, 64, initializer))]
elif name == 'ResBlock64_7':
links += [(name + '_' + str(i), ResBlock(7, 64, initializer))]
elif name == 'ResBlock128_3':
links += [(name + '_' + str(i), ResBlock(3, 128, initializer))]
elif name == 'ResBlock128_5':
links += [(name + '_' + str(i), ResBlock(5, 128, initializer))]
elif name == 'ResBlock128_7':
links += [(name + '_' + str(i), ResBlock(7, 128, initializer))]
elif name == 'full':
links += [(name + '_' + str(i),
L.Linear(None, n_class, initialW=initializer))]
i += 1
for link in links:
if not link[0].startswith('_'):
self.add_link(*link)
self.forward = links
self.train = True
self.lossfun = lossfun
self.accfun = accfun
self.loss = None
self.accuracy = None
self.outputs = [None for _ in range(len(self.cgp))]
self.param_num = 0