elif DATASET=='svhn':
dataset = sknet.dataset.load_svhn()
if "valid_set" not in dataset.sets:
dataset.split_set("train_set", "valid_set", 0.15)
standardize = sknet.dataset.Standardize().fit(dataset['images/train_set'])
dataset['images/train_set'] = \
standardize.transform(dataset['images/train_set'])
dataset['images/test_set'] = \
standardize.transform(dataset['images/test_set'])
dataset['images/valid_set'] = \
standardize.transform(dataset['images/valid_set'])
iterator = BatchIterator(32, {'train_set': "random_see_all",
'valid_set': 'continuous',
'test_set': 'continuous'})
dataset.create_placeholders(iterator, device="/cpu:0")
# Utility function
#-----------------
#c_p = tf.placeholder(tf.int32)
i_p = tf.placeholder(tf.int32)
j_p = tf.placeholder(tf.int32)
def get_distance(input,tensor):
def doit(c):
gradient = tf.gradients(tensor[:,c,i_p,j_p],input)[0]
norm = tf.sqrt(tf.reduce_sum(tf.square(gradient),[1,2,3]))
from sknet.dataset import BatchIterator
from sknet import ops, layers
from sknet.utils import flatten
from mpl_toolkits.mplot3d import Axes3D
import matplotlib.pyplot as plt
dataset = sknet.dataset.load_freefield1010(n_samples=2000, subsample=6)
dataset['signals/train_set'] /= dataset['signals/train_set'].max(2,
keepdims=True)
if "test_set" not in dataset.sets:
dataset.split_set("train_set", "test_set", 0.33)
dataset.create_placeholders(batch_size=5,
iterators_dict={
'train_set': BatchIterator("random_see_all"),
'valid_set': BatchIterator('continuous'),
'test_set': BatchIterator('continuous')
},
device="/cpu:0")
# Create Network
#---------------
dnn = sknet.Network(name='simple_model')
dnn.append(
sknet.ops.HermiteSplineConv1D(dataset.signals,
J=5,
Q=6,
K=15,
import matplotlib.pyplot as plt
import matplotlib.colors as mcolors
# Data Loading
#-------------
N = 500
TIME = np.linspace(-2, 2, N)
X = np.meshgrid(TIME, TIME)
X = np.stack([X[0].flatten(), X[1].flatten()], 1).astype('float32')
dataset = sknet.dataset.Dataset()
dataset.add_variable({'input': {'train_set': X}})
dataset.create_placeholders(
batch_size=50,
iterators_dict={'train_set': BatchIterator("continuous")},
device="/cpu:0")
# Create Network
#---------------
# we use a batch_size of 64 and use the dataset.datum shape to
# obtain the shape of 1 observation and create the input shape
# DN for the layer case
# RANK 1 PARALLEL
opt = int(sys.argv[-1])
b1 = np.asarray([-2.1, -1, -0.3, 1, 1.6, 2]).astype('float32') / 5
if opt == 0:
dnn = sknet.network.Network(name='simple_model')
# Data Loading
#-------------
dataset = sknet.dataset.load_cifar10()
dataset['images/train_set'] -= dataset['images/train_set'].mean((1, 2, 3),
keepdims=True)
dataset['images/train_set'] /= dataset['images/train_set'].max((1, 2, 3),
keepdims=True)
dataset['images/test_set'] -= dataset['images/test_set'].mean((1, 2, 3),
keepdims=True)
dataset['images/test_set'] /= dataset['images/test_set'].max((1, 2, 3),
keepdims=True)
iterator = BatchIterator(64, {
'train_set': 'random_see_all',
'test_set': 'continuous'
})
dataset.create_placeholders(iterator, device="/cpu:0")
# Create Network
#---------------
# we use a batch_size of 64 and use the dataset.datum shape to
# obtain the shape of 1 observation and create the input shape
dnn = sknet.Network(name='simple_model')
dnn.append(ops.RandomAxisReverse(dataset.images, axis=[-1]))
dnn.append(ops.RandomCrop(dnn[-1], (28, 28)))
elif DATASET=='svhn':
dataset = sknet.dataset.load_svhn()
if "valid_set" not in dataset.sets:
dataset.split_set("train_set","valid_set",0.15)
#standardize = sknet.dataset.Standardize().fit(dataset['images/train_set'])
#dataset['images/train_set'] = \
# standardize.transform(dataset['images/train_set'])
#dataset['images/test_set'] = \
# standardize.transform(dataset['images/test_set'])
#dataset['images/valid_set'] = \
# standardize.transform(dataset['images/valid_set'])
dataset.create_placeholders(batch_size=32,
iterators_dict={'train_set':BatchIterator("random_see_all"),
'valid_set':BatchIterator('continuous'),
'test_set':BatchIterator('continuous')},device="/cpu:0")
# Create Network
#---------------
dnn = sknet.Network(name='simple_model')
if DATA_AUGMENTATION:
dnn.append(ops.RandomAxisReverse(dataset.images,axis=[-1]))
dnn.append(ops.RandomCrop(dnn[-1],(28,28),seed=10))
else:
dnn.append(dataset.images)
if MODEL=='cnn':