def test_max_pooling2d(make_image_data):
"""Test dropout layer."""
x, _, X = make_image_data
# downsample by 2x
max_pool = ab.MaxPool2D(pool_size=(2, 2), strides=(2, 2))
F, KL = max_pool(X)
tc = tf.test.TestCase()
with tc.test_session():
f = F.eval()
# test equivalence of first window across batches
assert np.all(np.max(x[:, :2, :2, :], axis=(1, 2)) == f[0, :, 0, 0, :])
# n_samples and batch size remain unchanged
assert f.shape[:2] == X.eval().shape[:2]
# downsampled by 2x
assert 2 * f.shape[2] == X.eval().shape[2]
assert 2 * f.shape[3] == X.eval().shape[3]
# number of channels remain unchanged
assert f.shape[-1] == X.eval().shape[-1]
assert KL == 0
n_epochs = 50
batch_size = 100
config = tf.ConfigProto(device_count={'GPU': 0}) # Use GPU ?
reg = 0.1
l_samples = 5
p_samples = 5
# Network architecture
net = ab.stack(
ab.InputLayer(name='X', n_samples=l_samples), # LSAMPLES,BATCH_SIZE,28*28
ab.Conv2D(filters=32, kernel_size=(5, 5),
l2_reg=reg), # LSAMPLES, BATCH_SIZE, 28, 28, 32
ab.Activation(h=tf.nn.relu),
ab.MaxPool2D(pool_size=(2, 2),
strides=(2, 2)), # LSAMPLES, BATCH_SIZE, 14, 14, 32
ab.Conv2D(filters=64, kernel_size=(5, 5),
l2_reg=reg), # LSAMPLES, BATCH_SIZE, 14, 14, 64
ab.Activation(h=tf.nn.relu),
ab.MaxPool2D(pool_size=(2, 2),
strides=(2, 2)), # LSAMPLES, BATCH_SIZE, 7, 7, 64
ab.Flatten(), # LSAMPLES, BATCH_SIZE, 7*7*64
ab.Dense(output_dim=1024, l2_reg=reg), # LSAMPLES, BATCH_SIZE, 1024
ab.Activation(h=tf.nn.relu),
ab.DropOut(0.5),
ab.Dense(output_dim=10, l2_reg=reg), # LSAMPLES, BATCH_SIZE, 10
)
def main():