import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
import numpy as np
def make2D(vector, size=(101, 101)):
print vector.shape
return vector.reshape(size).T
def save_example_face(vector, fig_outfile):
im = make2D(vector)
plt.imshow(im, cmap='gray')
plt.tight_layout()
plt.savefig(fig_outfile, bbox_inches='tight')
if __name__ == "__main__":
from faces import faces
trX, trY, teX, teY = faces(contrast=True)
save_example_face(teX[0], 'example_face_contrast.png')
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
import numpy as np
def make2D(vector, size=(101, 101)):
print vector.shape
return vector.reshape(size).T
def save_example_face(vector, fig_outfile):
im = make2D(vector)
plt.imshow(im, cmap='gray')
plt.tight_layout()
plt.savefig(fig_outfile, bbox_inches='tight')
if __name__ == "__main__":
from faces import faces
trX, trY, teX, teY = faces(contrast=True)
save_example_face(teX[0], 'example_face_contrast.png')
import numpy as np
import theano
from six.moves import cPickle
from theano import tensor as T
from theano.sandbox.rng_mrg import MRG_RandomStreams as RandomStreams
from sklearn.metrics import accuracy_score
from faces import faces, permute
from performanceplot import performanceplot
import failure_analysis
srng = RandomStreams()
trX, trY, teX, teY = faces(zscore=False, onehot=False, adjust_targets=True, contrast=True)
print trX.shape
print trY.shape
print teX.shape
print teY.shape
input_dim = trX.shape[1]
mode = raw_input("What should this script do? (train, something else):")
def floatX(X):
return np.asarray(X, dtype=theano.config.floatX)
def init_weights(shape):
return theano.shared(floatX(np.random.randn(*shape) * 0.02))
def sgd(cost, params, lr=0.05):
grads = T.grad(cost=cost, wrt=params)
updates = []
for p, g in zip(params, grads):
import numpy as np
from foxhound.models import Network
from foxhound import ops
from foxhound import iterators
from foxhound.transforms import OneHot
from foxhound.theano_utils import floatX
from sklearn.metrics import accuracy_score
from faces import faces
binary_output = True
trX, trY, teX, teY = faces(zscore=True, adjust_targets=True)
def model_perceptron(input_shape):
model = [
ops.Input(['x', input_shape]),
ops.Project(dim=1),
ops.Activation('sigmoid')
]
return model
def model_MLP(input_shape):
model = [
ops.Input(['x', input_shape]),
ops.Project(dim=1000),
ops.Activation('tanh'),
import numpy as np
import theano
from theano import tensor as T
from theano.sandbox.rng_mrg import MRG_RandomStreams as RandomStreams
from sklearn.metrics import accuracy_score
from faces import faces, permute
from performanceplot import performanceplot
srng = RandomStreams()
trX, trY, teX, teY = faces(contrast=True, perceptron=True)
print "target values (min, max): ", (teY.min(), teY.max())
print trX.shape
print trY.shape
input_dim = trX.shape[1]
def floatX(X):
return np.asarray(X, dtype=theano.config.floatX)
def init_weights(shape):
return theano.shared(floatX(np.random.randn(*shape) * 0.02))
def model(X, w_o):
return T.sgn(T.dot(X, w_o))
X = T.fmatrix()
Y = T.fmatrix()
import numpy as np
import theano
from theano import tensor as T
from theano.sandbox.rng_mrg import MRG_RandomStreams as RandomStreams
from sklearn.metrics import accuracy_score
from faces import faces, permute
from performanceplot import performanceplot
srng = RandomStreams()
trX, trY, teX, teY = faces(contrast=True, perceptron=True)
print "target values (min, max): ", (teY.min(), teY.max())
print trX.shape
print trY.shape
input_dim = trX.shape[1]
def floatX(X):
return np.asarray(X, dtype=theano.config.floatX)
def init_weights(shape):
return theano.shared(floatX(np.random.randn(*shape) * 0.02))
def model(X, w_o):
return T.sgn(T.dot(X, w_o))
X = T.fmatrix()
Y = T.fmatrix()
# theano.config.compute_test_value = 'warn' # Use 'warn' to activate this featureg
# X.tag.test_value = np.zeros((1, input_dim), dtype='float32')
import numpy as np
import theano
from six.moves import cPickle
from theano import tensor as T
from theano.sandbox.rng_mrg import MRG_RandomStreams as RandomStreams
from sklearn.metrics import accuracy_score
from faces import faces, permute
from performanceplot import performanceplot
import failure_analysis
srng = RandomStreams()
trX, trY, teX, teY = faces(zscore=False,
onehot=False,
adjust_targets=True,
contrast=True)
print trX.shape
print trY.shape
print teX.shape
print teY.shape
input_dim = trX.shape[1]
mode = raw_input("What should this script do? (train, something else):")
def floatX(X):
return np.asarray(X, dtype=theano.config.floatX)
def init_weights(shape):
return theano.shared(floatX(np.random.randn(*shape) * 0.02))
import numpy as np
from foxhound.models import Network
from foxhound import ops
from foxhound import iterators
from foxhound.transforms import OneHot
from foxhound.theano_utils import floatX
from sklearn.metrics import accuracy_score
from faces import faces
binary_output = True
trX, trY, teX, teY = faces(zscore=True, adjust_targets=True)
def model_perceptron(input_shape):
model = [ops.Input(["x", input_shape]), ops.Project(dim=1), ops.Activation("sigmoid")]
return model
def model_MLP(input_shape):
model = [
ops.Input(["x", input_shape]),
ops.Project(dim=1000),
ops.Activation("tanh"),
ops.Project(dim=500),
ops.Activation("tanh"),
ops.Project(dim=75),
ops.Activation("tanh"),
