elif label=='jogging':
key = 4.
elif label=='running':
key = 5.
elif label=='walking':
key = 6.
else:
print "[ERROR] WRONG LABEL VALUE : ", label
temp = np.hstack((temp, key))
Yte=temp
del temp
print "[INFO] TRAIN DATA SHAPE : ", Xtr.shape
print "[INFO] TEST DATA SHAPE : ", Xte.shape
train_set_x, train_set_y=ds.shared_dataset((Xtr, Ytr));
test_set_x, test_set_y=ds.shared_dataset((Xte, Yte));
del Xtr, Xte, Ytr, Yte
n_train_batches=train_set_x.get_value(borrow=True).shape[0]
n_train_batches=int(np.ceil(n_train_batches/batch_size))
n_test_batches=test_set_x.get_value(borrow=True).shape[0]
n_test_batches=int(np.ceil(n_test_batches/batch_size))
print n_train_batches, n_test_batches
print "[MESSAGE] The data is loaded"
################################## LAYERWISE MODEL #######################################
X=T.matrix("data")
from scae_destin.cost import categorical_cross_entropy_cost
from scae_destin.cost import L2_regularization
start_time = time.time()
n_epochs = 100
batch_size = 100
nkerns = 100
Xtr, Ytr, Xte, Yte = ds.load_CIFAR10("../cifar-10-batches-py/")
Xtr = np.mean(Xtr, 3)
Xte = np.mean(Xte, 3)
Xtrain = Xtr.reshape(Xtr.shape[0], Xtr.shape[1] * Xtr.shape[2]) / 255.0
Xtest = Xte.reshape(Xte.shape[0], Xte.shape[1] * Xte.shape[2]) / 255.0
train_set_x, train_set_y = ds.shared_dataset((Xtrain, Ytr))
test_set_x, test_set_y = ds.shared_dataset((Xtest, Yte))
n_train_batches = train_set_x.get_value(borrow=True).shape[0] / batch_size
n_test_batches = test_set_x.get_value(borrow=True).shape[0] / batch_size
print "[MESSAGE] The data is loaded"
################################## FIRST LAYER #######################################
X = T.matrix("data")
y = T.ivector("label")
idx = T.lscalar()
corruption_level = T.fscalar()
images = X.reshape((batch_size, 1, 32, 32))
from scae_destin.cost import L2_regularization
#def conv():
n_epochs=1
batch_size=100
nkerns=100
Xtr, Ytr, Xte, Yte=ds.load_CIFAR10("../cifar-10-batches-py/")
Xtr=np.mean(Xtr, 3)
Xte=np.mean(Xte, 3)
Xtrain=Xtr.reshape(Xtr.shape[0], Xtr.shape[1]*Xtr.shape[2])/255.0
Xtest=Xte.reshape(Xte.shape[0], Xte.shape[1]*Xte.shape[2])/255.0
train_set_x, train_set_y=ds.shared_dataset((Xtrain, Ytr))
test_set_x, test_set_y=ds.shared_dataset((Xtest, Yte))
n_train_batches=train_set_x.get_value(borrow=True).shape[0]/batch_size
n_test_batches=test_set_x.get_value(borrow=True).shape[0]/batch_size
print "[MESSAGE] The data is loaded"
################################## FIRST LAYER #######################################
X=T.matrix("data")
y=T.ivector("label")
idx=T.lscalar()
corruption_level=T.fscalar()
images=X.reshape((batch_size, 1, 32, 32))
