def setUp(self):
skip_if_no_h5py()
import h5py
skip_if_no_data()
from pylearn2.datasets.mnist import MNIST
# save MNIST data to HDF5
train = MNIST(which_set='train', one_hot=1, start=0, stop=100)
for name, dataset in [('train', train)]:
with h5py.File("{}.h5".format(name), "w") as f:
f.create_dataset('X', data=dataset.get_design_matrix())
f.create_dataset('topo_view',
data=dataset.get_topological_view())
f.create_dataset('y', data=dataset.get_targets())
# instantiate Train object
self.train = yaml_parse.load(trainer_yaml)
def setUp(self):
skip_if_no_h5py()
import h5py
skip_if_no_data()
from pylearn2.datasets.mnist import MNIST
# save MNIST data to HDF5
train = MNIST(which_set='train', one_hot=1, start=0, stop=100)
for name, dataset in [('train', train)]:
with h5py.File("{}.h5".format(name), "w") as f:
f.create_dataset('X', data=dataset.get_design_matrix())
f.create_dataset('topo_view',
data=dataset.get_topological_view())
f.create_dataset('y', data=dataset.get_targets())
# instantiate Train object
self.train = yaml_parse.load(trainer_yaml)
def test_hdf5_convert_to_one_hot():
"""Train using an HDF5 dataset with one-hot target conversion."""
skip_if_no_h5py()
import h5py
# save random data to HDF5
handle, filename = tempfile.mkstemp()
dataset = random_dense_design_matrix(np.random.RandomState(1),
num_examples=10, dim=5, num_classes=3)
with h5py.File(filename, 'w') as f:
f.create_dataset('X', data=dataset.get_design_matrix())
f.create_dataset('y', data=dataset.get_targets())
# instantiate Train object
trainer = yaml_parse.load(convert_to_one_hot_yaml % {'filename': filename})
trainer.main_loop()
# cleanup
os.remove(filename)
def test_hdf5_topo_view():
"""Train using an HDF5 dataset with topo_view instead of X."""
skip_if_no_h5py()
import h5py
# save random data to HDF5
handle, filename = tempfile.mkstemp()
dataset = random_one_hot_topological_dense_design_matrix(
np.random.RandomState(1), num_examples=10, shape=(2, 2), channels=3,
axes=('b', 0, 1, 'c'), num_classes=3)
with h5py.File(filename, 'w') as f:
f.create_dataset('topo_view', data=dataset.get_topological_view())
f.create_dataset('y', data=dataset.get_targets())
# instantiate Train object
trainer = yaml_parse.load(topo_view_yaml % {'filename': filename})
trainer.main_loop()
# cleanup
os.remove(filename)
def test_hdf5_load_all():
"""Train using an HDF5 dataset with all data loaded into memory."""
skip_if_no_h5py()
import h5py
# save random data to HDF5
handle, filename = tempfile.mkstemp()
dataset = random_one_hot_dense_design_matrix(np.random.RandomState(1),
num_examples=10,
dim=5,
num_classes=3)
with h5py.File(filename, 'w') as f:
f.create_dataset('X', data=dataset.get_design_matrix())
f.create_dataset('y', data=dataset.get_targets())
# instantiate Train object
trainer = yaml_parse.load(load_all_yaml % {'filename': filename})
trainer.main_loop()
# cleanup
os.remove(filename)
def test_hdf5_topo_view():
"""Train using an HDF5 dataset with topo_view instead of X."""
skip_if_no_h5py()
import h5py
# save random data to HDF5
handle, filename = tempfile.mkstemp()
dataset = random_one_hot_topological_dense_design_matrix(
np.random.RandomState(1),
num_examples=10,
shape=(2, 2),
channels=3,
axes=('b', 0, 1, 'c'),
num_classes=3)
with h5py.File(filename, 'w') as f:
f.create_dataset('topo_view', data=dataset.get_topological_view())
f.create_dataset('y', data=dataset.get_targets())
# instantiate Train object
trainer = yaml_parse.load(topo_view_yaml % {'filename': filename})
trainer.main_loop()
# cleanup
os.remove(filename)