使用Sigmoid做为激活函数的两层网络，如图14-12。
def model_sigmoid(num_input, num_hidden, num_output, hp):
net = NeuralNet_4_0(hp, "chinabank_sigmoid")
fc1 = FcLayer_1_0(num_input, num_hidden, hp)
net.add_layer(fc1, "fc1")
s1 = ActivationLayer(Sigmoid())
net.add_layer(s1, "Sigmoid1")
fc2 = FcLayer_1_0(num_hidden, num_output, hp)
net.add_layer(fc2, "fc2")
softmax1 = ClassificationLayer(Softmax())
net.add_layer(softmax1, "softmax1")
net.train(dataReader, checkpoint=50, need_test=True)
net.ShowLossHistory()
ShowResult(net, hp.toString())
ShowData(dataReader)- 隐层8个神经元
- 最大epoch=5000
- 批大小=10
- 学习率0.1
- 绝对误差停止条件=0.08
- 多分类网络类型
- 初始化方法为Xavier
net.train()函数是一个阻塞函数，只有当训练完毕后才返回。
训练过程如图14-13所示，分类效果如图14-14所示。
使用ReLU做为激活函数的三层网络，如图14-15。
用两层网络也可以实现，但是使用ReLE函数时，训练效果不是很稳定，用三层比较保险。
def model_relu(num_input, num_hidden, num_output, hp):
net = NeuralNet_4_0(hp, "chinabank_relu")
fc1 = FcLayer_1_0(num_input, num_hidden, hp)
net.add_layer(fc1, "fc1")
r1 = ActivationLayer(Relu())
net.add_layer(r1, "Relu1")
fc2 = FcLayer_1_0(num_hidden, num_hidden, hp)
net.add_layer(fc2, "fc2")
r2 = ActivationLayer(Relu())
net.add_layer(r2, "Relu2")
fc3 = FcLayer_1_0(num_hidden, num_output, hp)
net.add_layer(fc3, "fc3")
softmax = ClassificationLayer(Softmax())
net.add_layer(softmax, "softmax")
net.train(dataReader, checkpoint=50, need_test=True)
net.ShowLossHistory()
ShowResult(net, hp.toString())
ShowData(dataReader) - 隐层8个神经元
- 最大epoch=5000
- 批大小=10
- 学习率0.1
- 绝对误差停止条件=0.08
- 多分类网络类型
- 初始化方法为MSRA
训练过程如图14-16所示，分类效果如图14-17所示。
表14-1比较一下使用不同的激活函数的分类效果图。
表14-1 使用不同的激活函数的分类结果比较
| Sigmoid | ReLU |
|---|---|
|
|
可以看到左图中的边界要平滑许多，这也就是ReLU和Sigmoid的区别，ReLU是用分段线性拟合曲线，Sigmoid有真正的曲线拟合能力。但是Sigmoid也有缺点，看分类的边界，使用ReLU函数的分类边界比较清晰，而使用Sigmoid函数的分类边界要平缓一些，过渡区较宽。
用一句简单的话来描述二者的差别：Relu能直则直，对方形边界适用；Sigmoid能弯则弯，对圆形边界适用。
原代码位置：ch14, Level5
个人代码：dnn_multiClassification****
from MiniFramework.DataReader_2_0 import *
from keras.models import Sequential
from keras.layers import Dense
import matplotlib.pyplot as plt
import os
os.environ['KMP_DUPLICATE_LIB_OK']='True'
def load_data():
train_file = "../data/ch11.train.npz"
test_file = "../data/ch11.test.npz"
dataReader = DataReader_2_0(train_file, test_file)
dataReader.ReadData()
dataReader.NormalizeX()
dataReader.NormalizeY(NetType.MultipleClassifier, base=1)
dataReader.Shuffle()
dataReader.GenerateValidationSet()
x_train, y_train = dataReader.XTrain, dataReader.YTrain
x_test, y_test = dataReader.XTest, dataReader.YTest
x_val, y_val = dataReader.XDev, dataReader.YDev
return x_train, y_train, x_test, y_test, x_val, y_val
def build_model():
model = Sequential()
model.add(Dense(8, activation='relu', input_shape=(2, )))
model.add(Dense(8, activation='relu'))
model.add(Dense(3, activation='softmax'))
model.compile(optimizer='Adam',
loss='categorical_crossentropy',
metrics=['accuracy'])
return model
#画出训练过程中训练和验证的精度与损失
def draw_train_history(history):
plt.figure(1)
# summarize history for accuracy
plt.subplot(211)
plt.plot(history.history['accuracy'])
plt.plot(history.history['val_accuracy'])
plt.title('model accuracy')
plt.ylabel('accuracy')
plt.xlabel('epoch')
plt.legend(['train', 'validation'])
# summarize history for loss
plt.subplot(212)
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title('model loss')
plt.ylabel('loss')
plt.xlabel('epoch')
plt.legend(['train', 'validation'])
plt.show()
if __name__ == '__main__':
x_train, y_train, x_test, y_test, x_val, y_val = load_data()
model = build_model()
history = model.fit(x_train, y_train, epochs=100, batch_size=10, validation_data=(x_val, y_val))
draw_train_history(history)
loss, accuracy = model.evaluate(x_test, y_test)
print("test loss: {}, test accuracy: {}".format(loss, accuracy))
weights = model.get_weights()
print("weights: ", weights)模型输出
test loss: 0.17712581551074982, test accuracy: 0.9480000138282776
weights: [array([[ 0.5933495 , 1.4001606 , -0.46128672, -1.0328066 , 0.4704834 ,
-0.13881624, -1.5763694 , 0.39080548],
[-1.368007 , 0.6859568 , -0.73229355, 0.7306851 , 0.7245091 ,
0.07994052, -0.24856903, -1.7596178 ]], dtype=float32), array([ 0.45760065, -0.15192626, 0. , 0.07134774, -0.5471344 ,
0.8713162 , 0.70304626, 0.35331267], dtype=float32), array([[-0.44358087, 1.4647273 , 0.6838032 , 1.3377259 , 0.9623304 ,
0.801778 , 0.07605773, 0.3418852 ],
[-0.4460683 , 0.34761527, 1.0513452 , -0.06174321, -1.0383745 ,
0.6795882 , 0.10854041, 0.7523184 ],
[-0.10631716, 0.11873323, -0.6039761 , 0.25613695, -0.52250814,
-0.30054256, -0.21584505, -0.1406537 ],
[-0.16741991, 1.4979596 , 2.0585673 , 1.7580718 , -0.12573877,
1.0612497 , 0.3230644 , -0.5291618 ],
[-0.5735318 , 1.6553403 , 1.7515256 , 2.3439772 , -0.6199714 ,
1.8710839 , -0.479978 , 0.32344452],
[ 0.05696827, -0.42794174, -0.84942245, 0.140646 , 0.10621271,
-0.6504364 , -0.46572435, 1.5581474 ],
[-0.5257491 , 3.3067589 , 0.88320696, 3.685748 , 1.4463454 ,
1.6930596 , -0.4242273 , 0.01312767],
[-0.17542031, 2.562144 , 0.5151277 , 2.2590203 , 0.89615613,
0.673661 , -0.22245789, -0.5107478 ]], dtype=float32), array([ 0. , -0.279784 , -0.7419632 , -0.40435618, -0.25443217,
-0.5012585 , -0.05188801, 1.19713 ], dtype=float32), array([[-0.70594245, -0.10075217, -0.66060597],
[-4.915914 , 0.15276138, 2.171216 ],
[-4.7157874 , -1.9014189 , 2.9048522 ],
[-4.163957 , 0.42147762, 1.3650419 ],
[-2.2925568 , -2.7893898 , 3.548446 ],
[-4.951117 , -1.8089052 , 2.7878227 ],
[-0.04061006, -0.04594503, -0.29654002],
[ 1.1871618 , 0.6122024 , -1.6027299 ]], dtype=float32), array([ 1.4912351 , 0.29161552, -1.5535753 ], dtype=float32)]
Process finished with exit code 0模型损失以及准确率曲线
