def CNNForward(model, x):
"""Runs the forward pass.
Args:
model: Dictionary of all the weights.
x: Input to the network.
Returns:
var: Dictionary of all intermediate variables.
"""
x = x.reshape([-1, 48, 48, 1])
h1c = Conv2D(x, model['W1']) + model['b1']
h1r = ReLU(h1c)
h1p = MaxPool(h1r, 3)
h2c = Conv2D(h1p, model['W2']) + model['b2']
h2r = ReLU(h2c)
h2p = MaxPool(h2r, 2)
h2p_ = np.reshape(h2p, [x.shape[0], -1])
y = Affine(h2p_, model['W3'], model['b3'])
var = {
'x': x,
'h1c': h1c,
'h1r': h1r,
'h1p': h1p,
'h2c': h2c,
'h2r': h2r,
'h2p': h2p,
'h2p_': h2p_,
'y': y
}
return var
def gen(Z, w, w1, w2, w3):
h0 = ReLU(batchnorm(T.dot(Z, w)))
h1 = ReLU(batchnorm(T.dot(h0, w1)))
h1_output = h1.reshape((h1.shape[0], nkerns[2], 7, 7))
h2_input = repeat(repeat(h1_output, 2, 2), 2, 3)
h2 = ReLU(batchnorm(conv2d(h2_input, w2, border_mode='half')))
h3_input = repeat(repeat(h2, 2, 2), 2, 3)
h3 = T.tanh(conv2d(h3_input, w3, border_mode='half'))
return h3
def __init__(self):
self.layers = [
Conv2d(1, 6, 5, padding=2),
ReLU(),
MaxPool2d(2, 2),
Conv2d(6, 16, 5),
ReLU(),
MaxPool2d(2, 2),
Flatten(),
Linear(400, 120),
ReLU(),
Linear(120, 84),
ReLU(),
Linear(84, 10)
]
def gen(Z, w, w1, w2, w3, w4):
h0 = ReLU(batchnorm(T.dot(Z, w)))
h1_input = h0.reshape((h0.shape[0], nkerns[3], 4, 4))
h1 = ReLU(batchnorm(conv2d(h1_input, w1, border_mode='half')))
h2_input = repeat(repeat(h1, 2, 2), 2, 3)
h2 = ReLU(batchnorm(conv2d(h2_input, w2, border_mode='half')))
h3_input = repeat(repeat(h2, 2, 2), 2, 3)
h3 = ReLU(batchnorm(conv2d(h3_input, w3, border_mode='half')))
h4_input = repeat(repeat(h3, 2, 2), 2, 3)
h4 = T.tanh(conv2d(h4_input, w4, border_mode='half'))
return h4