givens={
x: train_set_x[index:index+1]})
print get_example_th(0)
print type(get_example_th(1))
print get_example_th(2)[0]
from visualize import *
print get_example_th(0)[0].shape
show_4_ex(get_example_th(0)[0].reshape(ImageChannels, ImageSideFinal, ImageSideFinal))
#############
# Train Model #
##############
# compute number of minibatches for training, validation and testing
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 '... training the model'
# early-stopping parameters
patience = 5000 # look as this many examples regardless
patience_increase = 2 # wait this much longer when a new best is
import cPickle
cPickle.dump(model, open(MODEL_NAME,"w"))
else:
layer0, layer1, layer2, layer3 = cPickle.load(open(MODEL_NAME, "r"))
from visualize import show_4_ex
from utils import tile_raster_images
import matplotlib.pylab as plt
#plt.imshow(tile_raster_images(
# X=layer0.W.get_value(borrow=True).T,
# img_shape=(data_api.ImageSideFinal, data_api.ImageSideFinal), tile_shape=(10, 10),
# tile_spacing=(1, 1)))
#plt.show()
show_4_ex(layer0.W.get_value()[0])
show_4_ex(layer0.W.get_value()[1])
try:
for i in xrange(10000):
xy_net = get_example_th_tst(i)
x_net = xy_net[0][0]
y_net = xy_net[1]
if (logReg and y_net[0]==0) or (not logReg and y_net[0,0] == 0):
print "0 example"
print predict_test(i)
error_net = test_model(i)
plt.imshow(x_net.reshape(ImageChannels, ImageSideFinal, ImageSideFinal)[0])
plt.colorbar()
plt.title("Label "+str(y_net)+" Error " + str(error_net)+ " Activation = " +str(predict_test(i)[0]))
plt.show()