def dnn_model():
num_output = 9
max_epoch = 50
batch_size = 16
learning_rate = 0.01
params = HyperParameters_4_2(learning_rate,
max_epoch,
batch_size,
net_type=NetType.MultipleClassifier,
init_method=InitialMethod.MSRA,
optimizer_name=OptimizerName.Momentum)
net = NeuralNet_4_2(params, "color_shape_dnn")
f1 = FcLayer_2_0(784, 128, params)
net.add_layer(f1, "f1")
r1 = ActivationLayer(Relu())
net.add_layer(r1, "relu1")
f2 = FcLayer_2_0(f1.output_size, 64, params)
net.add_layer(f2, "f2")
r2 = ActivationLayer(Relu())
net.add_layer(r2, "relu2")
f3 = FcLayer_2_0(f2.output_size, num_output, params)
net.add_layer(f3, "f3")
s3 = ClassificationLayer(Softmax())
net.add_layer(s3, "s3")
return net
def Model(dataReader, num_input, num_hidden, num_output, params):
net = NeuralNet_4_2(params, "overfitting")
fc1 = FcLayer_2_0(num_input, num_hidden, params)
net.add_layer(fc1, "fc1")
s1 = ActivatorLayer(Sigmoid())
net.add_layer(s1, "s1")
fc2 = FcLayer_2_0(num_hidden, num_hidden, params)
net.add_layer(fc2, "fc2")
t2 = ActivatorLayer(Tanh())
net.add_layer(t2, "t2")
fc3 = FcLayer_2_0(num_hidden, num_hidden, params)
net.add_layer(fc3, "fc3")
t3 = ActivatorLayer(Tanh())
net.add_layer(t3, "t3")
fc4 = FcLayer_2_0(num_hidden, num_output, params)
net.add_layer(fc4, "fc4")
net.train(dataReader, checkpoint=100, need_test=True)
net.ShowLossHistory(XCoordinate.Epoch)
return net
def model():
num_output = 10
dataReader = LoadData()
max_epoch = 2
batch_size = 128
learning_rate = 0.1
params = HyperParameters_4_2(learning_rate,
max_epoch,
batch_size,
net_type=NetType.MultipleClassifier,
init_method=InitialMethod.MSRA,
optimizer_name=OptimizerName.Momentum)
net = NeuralNet_4_2(params, "cifar_conv")
c1 = ConvLayer((3, 32, 32), (6, 5, 5), (1, 0), params)
net.add_layer(c1, "c1")
r1 = ActivationLayer(Relu())
net.add_layer(r1, "relu1")
p1 = PoolingLayer(c1.output_shape, (2, 2), 2, PoolingTypes.MAX)
net.add_layer(p1, "p1")
c2 = ConvLayer(p1.output_shape, (16, 5, 5), (1, 0), params)
net.add_layer(c2, "c2")
r2 = ActivationLayer(Relu())
net.add_layer(r2, "relu2")
p2 = PoolingLayer(c2.output_shape, (2, 2), 2, PoolingTypes.MAX)
net.add_layer(p2, "p2")
f1 = FcLayer_2_0(p2.output_size, 120, params)
net.add_layer(f1, "f1")
bn1 = BnLayer(120)
net.add_layer(bn1, "bn1")
r3 = ActivationLayer(Relu())
net.add_layer(r3, "relu3")
f2 = FcLayer_2_0(f1.output_size, 84, params)
net.add_layer(f2, "f2")
bn2 = BnLayer(84)
net.add_layer(bn2, "bn2")
r4 = ActivationLayer(Relu())
net.add_layer(r4, "relu4")
f3 = FcLayer_2_0(f2.output_size, num_output, params)
net.add_layer(f3, "f3")
s1 = ClassificationLayer(Softmax())
net.add_layer(s1, "s1")
#net.load_parameters()
net.train(dataReader, checkpoint=0.05, need_test=True)
net.ShowLossHistory(XCoordinate.Iteration)
def model():
num_output = 10
max_epoch = 10
batch_size = 32
learning_rate = 0.1
params = HyperParameters_4_2(learning_rate,
max_epoch,
batch_size,
net_type=NetType.MultipleClassifier,
init_method=InitialMethod.MSRA,
optimizer_name=OptimizerName.Momentum)
net = NeuralNet_4_2(params, "cifar_conv")
c1 = ConvLayer((3, 32, 32), (32, 3, 3), (1, 1), params)
net.add_layer(c1, "c1")
r1 = ActivationLayer(Relu())
net.add_layer(r1, "relu1")
c2 = ConvLayer(c1.output_shape, (32, 3, 3), (1, 0), params)
net.add_layer(c2, "c2")
r2 = ActivationLayer(Relu())
net.add_layer(r2, "relu2")
p2 = PoolingLayer(c2.output_shape, (2, 2), 2, PoolingTypes.MAX)
net.add_layer(p2, "p2")
c3 = ConvLayer(p2.output_shape, (64, 3, 3), (1, 1), params)
net.add_layer(c3, "c3")
r3 = ActivationLayer(Relu())
net.add_layer(r3, "relu3")
c4 = ConvLayer(c3.output_shape, (64, 3, 3), (1, 0), params)
net.add_layer(c4, "c4")
r4 = ActivationLayer(Relu())
net.add_layer(r4, "relu4")
p4 = PoolingLayer(c4.output_shape, (2, 2), 2, PoolingTypes.MAX)
net.add_layer(p4, "p4")
f5 = FcLayer_2_0(p4.output_size, 512, params)
net.add_layer(f5, "f5")
r5 = ActivationLayer(Relu())
net.add_layer(r5, "relu5")
f6 = FcLayer_2_0(f5.output_size, num_output, params)
net.add_layer(f6, "f6")
s7 = ClassificationLayer(Softmax())
net.add_layer(s7, "s7")
return net
def model():
num_output = 10
max_epoch = 5
batch_size = 64
learning_rate = 0.1
params = HyperParameters_4_2(
learning_rate, max_epoch, batch_size,
net_type=NetType.MultipleClassifier,
init_method=InitialMethod.MSRA,
optimizer_name=OptimizerName.Momentum)
net = NeuralNet_4_2(params, "cifar_conv")
c1 = ConvLayer((3,32,32), (6,5,5), (1,0), params)
net.add_layer(c1, "c1")
r1 = ActivationLayer(Relu())
net.add_layer(r1, "relu1")
p1 = PoolingLayer(c1.output_shape, (2,2), 2, PoolingTypes.MAX)
net.add_layer(p1, "p1")
c2 = ConvLayer(p1.output_shape, (16,5,5), (1,0), params)
net.add_layer(c2, "c2")
r2 = ActivationLayer(Relu())
net.add_layer(r2, "relu2")
p2 = PoolingLayer(c2.output_shape, (2,2), 2, PoolingTypes.MAX)
net.add_layer(p2, "p2")
f3 = FcLayer_2_0(p2.output_size, 120, params)
net.add_layer(f3, "f3")
bn3 = BnLayer(120)
net.add_layer(bn3, "bn3")
r3 = ActivationLayer(Relu())
net.add_layer(r3, "relu3")
f4 = FcLayer_2_0(f3.output_size, 84, params)
net.add_layer(f4, "f4")
bn4 = BnLayer(84)
net.add_layer(bn4, "bn4")
r4 = ActivationLayer(Relu())
net.add_layer(r4, "relu4")
f5 = FcLayer_2_0(f4.output_size, num_output, params)
net.add_layer(f5, "f5")
s5 = ClassificationLayer(Softmax())
net.add_layer(s5, "s5")
return net
def model():
num_output = 10
dataReader = LoadData(num_output)
max_epoch = 5
batch_size = 128
learning_rate = 0.1
params = HyperParameters_4_2(learning_rate,
max_epoch,
batch_size,
net_type=NetType.MultipleClassifier,
init_method=InitialMethod.Xavier,
optimizer_name=OptimizerName.Momentum)
net = NeuralNet_4_2(params, "mnist_conv_test")
c1 = ConvLayer((1, 28, 28), (8, 5, 5), (1, 0), params)
net.add_layer(c1, "c1")
r1 = ActivationLayer(Relu())
net.add_layer(r1, "relu1")
p1 = PoolingLayer(c1.output_shape, (2, 2), 2, PoolingTypes.MAX)
net.add_layer(p1, "p1")
c2 = ConvLayer(p1.output_shape, (16, 5, 5), (1, 0), params)
net.add_layer(c2, "23")
r2 = ActivationLayer(Relu())
net.add_layer(r2, "relu2")
p2 = PoolingLayer(c2.output_shape, (2, 2), 2, PoolingTypes.MAX)
net.add_layer(p2, "p2")
f3 = FcLayer_2_0(p2.output_size, 32, params)
net.add_layer(f3, "f3")
bn3 = BnLayer(f3.output_size)
net.add_layer(bn3, "bn3")
r3 = ActivationLayer(Relu())
net.add_layer(r3, "relu3")
f4 = FcLayer_2_0(f3.output_size, 10, params)
net.add_layer(f4, "f2")
s4 = ClassificationLayer(Softmax())
net.add_layer(s4, "s4")
net.train(dataReader, checkpoint=0.05, need_test=True)
net.ShowLossHistory(XCoordinate.Iteration)
def cnn_model():
num_output = 6
max_epoch = 20
batch_size = 16
learning_rate = 0.1
params = HyperParameters_4_2(
learning_rate, max_epoch, batch_size,
net_type=NetType.MultipleClassifier,
init_method=InitialMethod.MSRA,
optimizer_name=OptimizerName.SGD)
net = NeuralNet_4_2(params, "color_cnn")
c1 = ConvLayer((3,28,28), (2,1,1), (1,0), params)
net.add_layer(c1, "c1")
r1 = ActivationLayer(Relu())
net.add_layer(r1, "relu1")
p1 = PoolingLayer(c1.output_shape, (2,2), 2, PoolingTypes.MAX)
net.add_layer(p1, "p1")
"""
c2 = ConvLayer(p1.output_shape, (3,3,3), (1,0), params)
net.add_layer(c2, "c2")
r2 = ActivationLayer(Relu())
net.add_layer(r2, "relu2")
p2 = PoolingLayer(c2.output_shape, (2,2), 2, PoolingTypes.MAX)
net.add_layer(p2, "p2")
"""
params.learning_rate = 0.1
#f3 = FcLayer_2_0(p2.output_size, 32, params)
f3 = FcLayer_2_0(p1.output_size, 32, params)
net.add_layer(f3, "f3")
bn3 = BnLayer(f3.output_size)
net.add_layer(bn3, "bn3")
r3 = ActivationLayer(Relu())
net.add_layer(r3, "relu3")
f4 = FcLayer_2_0(f3.output_size, num_output, params)
net.add_layer(f4, "f4")
s4 = ClassificationLayer(Softmax())
net.add_layer(s4, "s4")
return net
def model():
num_output = 10
dataReader = LoadData(num_output)
max_epoch = 5
batch_size = 128
learning_rate = 0.1
params = HyperParameters_4_2(
learning_rate, max_epoch, batch_size,
net_type=NetType.MultipleClassifier,
init_method=InitialMethod.MSRA,
optimizer_name=OptimizerName.Momentum)
net = NeuralNet_4_2(params, "mnist_conv")
c1 = ConvLayer((1,28,28), (8,3,3), (1,1), params)
net.add_layer(c1, "c1")
r1 = ActivationLayer(Relu())
net.add_layer(r1, "relu1")
p1 = PoolingLayer(c1.output_shape, (2,2), 2, PoolingTypes.MAX)
net.add_layer(p1, "p1")
c3 = ConvLayer(p1.output_shape, (16,3,3), (1,1), params)
net.add_layer(c3, "c3")
r3 = ActivationLayer(Relu())
net.add_layer(r3, "relu3")
p2 = PoolingLayer(c3.output_shape, (2,2), 2, PoolingTypes.MAX)
net.add_layer(p2, "p2")
f1 = FcLayer_2_0(p2.output_size, 32, params)
net.add_layer(f1, "f1")
bn1 = BnLayer(f1.output_size)
net.add_layer(bn1, "bn1")
r5 = ActivationLayer(Relu())
net.add_layer(r5, "relu5")
f2 = FcLayer_2_0(f1.output_size, 10, params)
net.add_layer(f2, "f2")
s1 = ClassificationLayer(Softmax())
net.add_layer(s1, "s1")
net.train(dataReader, checkpoint=0.01, need_test=True)
net.ShowLossHistory(XCoordinate.Iteration)
def model():
num_output = 10
max_epoch = 5
batch_size = 128
learning_rate = 0.1
params = HyperParameters_4_2(learning_rate,
max_epoch,
batch_size,
net_type=NetType.MultipleClassifier,
init_method=InitialMethod.Xavier,
optimizer_name=OptimizerName.Momentum)
net = NeuralNet_4_2(params, "mnist_cnn")
c1 = ConvLayer((1, 28, 28), (8, 5, 5), (1, 0), params)
net.add_layer(c1, "c1")
r1 = ActivationLayer(Relu())
net.add_layer(r1, "relu1")
p1 = PoolingLayer(c1.output_shape, (2, 2), 2, PoolingTypes.MAX)
net.add_layer(p1, "p1")
c2 = ConvLayer(p1.output_shape, (16, 5, 5), (1, 0), params)
net.add_layer(c2, "c2")
r2 = ActivationLayer(Relu())
net.add_layer(r2, "relu2")
p2 = PoolingLayer(c2.output_shape, (2, 2), 2, PoolingTypes.MAX)
net.add_layer(p2, "p2")
f3 = FcLayer_2_0(p2.output_size, 32, params)
net.add_layer(f3, "f3")
bn3 = BnLayer(f3.output_size)
net.add_layer(bn3, "bn3")
r3 = ActivationLayer(Relu())
net.add_layer(r3, "relu3")
f4 = FcLayer_2_0(f3.output_size, 10, params)
net.add_layer(f4, "f2")
s4 = ClassificationLayer(Softmax())
net.add_layer(s4, "s4")
return net
def Net(subfolder,
dataReader,
num_input,
num_hidden,
num_output,
params,
show_history=True):
net = NeuralNet_4_2(params, subfolder)
fc1 = FcLayer_2_0(num_input, num_hidden, params)
net.add_layer(fc1, "fc1")
relu1 = ActivatorLayer(Relu())
net.add_layer(relu1, "relu1")
fc2 = FcLayer_2_0(num_hidden, num_hidden, params)
net.add_layer(fc2, "fc2")
relu2 = ActivatorLayer(Relu())
net.add_layer(relu2, "relu2")
fc3 = FcLayer_2_0(num_hidden, num_hidden, params)
net.add_layer(fc3, "fc3")
relu3 = ActivatorLayer(Relu())
net.add_layer(relu3, "relu3")
fc4 = FcLayer_2_0(num_hidden, num_hidden, params)
net.add_layer(fc4, "fc4")
relu4 = ActivatorLayer(Relu())
net.add_layer(relu4, "relu4")
fc5 = FcLayer_2_0(num_hidden, num_output, params)
net.add_layer(fc5, "fc5")
softmax = ClassificationLayer(Softmax())
net.add_layer(softmax, "softmax")
net.train(dataReader, checkpoint=1, need_test=True)
if show_history:
net.ShowLossHistory(XCoordinate.Iteration)
return net
