def test_dropout(self):
np.random.seed(1)
x = np.random.randn(3, 5)
w1 = np.random.randn(2, 3)
b1 = np.random.randn(2, 1).reshape(1, 2)
w2 = np.random.randn(3, 2)
b2 = np.random.randn(3, 1).reshape(1, 3)
w3 = np.random.randn(1, 3)
b3 = np.random.randn(1, 1)
nn = NeuralNetwork(
input_dim=3,
layers=[
FullyConnected(2, relu),
Dropout(keep_prob=0.7),
FullyConnected(3, relu),
Dropout(keep_prob=0.7),
FullyConnected(1, sigmoid)
],
cost_function=sigmoid_cross_entropy,
)
nn.layers[0].w = w1
nn.layers[0].b = b1
nn.layers[2].w = w2
nn.layers[2].b = b2
nn.layers[4].w = w3
nn.layers[4].b = b3
np.random.seed(1)
a_last = nn.forward_prop(x.T)
np.testing.assert_array_almost_equal(
a_last,
np.array([[0.369747, 0.496834, 0.045651, 0.014469, 0.369747]]).T)
def test_forward(self):
np.random.seed(6)
x = np.random.randn(5, 4)
x_num = 5
h1_num = 4
h2_num = 3
y_num = 1
w1 = np.random.randn(h1_num, x_num)
b1 = np.random.randn(1, h1_num)
w2 = np.random.randn(h2_num, h1_num)
b2 = np.random.randn(1, h2_num)
w3 = np.random.randn(y_num, h2_num)
b3 = np.random.randn(1, y_num)
nn = NeuralNetwork(input_dim=x_num,
layers=[
FullyConnected(h1_num, relu),
FullyConnected(h2_num, relu),
FullyConnected(y_num, sigmoid)
],
cost_function=sigmoid_cross_entropy)
nn.layers[0].w = w1
nn.layers[0].b = b1
nn.layers[1].w = w2
nn.layers[1].b = b2
nn.layers[2].w = w3
nn.layers[2].b = b3
a_last = nn.forward_prop(x.T)
np.testing.assert_array_almost_equal(
a_last,
np.array([[0.03921668, 0.70498921, 0.19734387, 0.04728177]]).T)