西瓜书第五章笔记（待补）

5.1 神经元模型
5.2 感知机与多层网络
5.3 BP神经网络算法
5.4 全局最小与局部最小
5.5 深度学习
入门例子：手写数字识别，使用mnist数据集进行训练，代码如下：

from keras.datasets.mnist import load_data
from keras.utils.np_utils import *
from keras import Sequential
from keras.layers import Dense, Activation, Dropout
from keras.layers import Conv2D, MaxPooling2D, Flatten, Conv1D,MaxPooling1D
from keras.optimizers import SGD, Adam
from keras.models import load_model
from PIL import Image
import numpy as np
import keras


def img_show(img):
    pil_img = Image.fromarray(np.uint8(img))
    pil_img.show()

if __name__ == '__main__':
    '''数据读入部分'''
    (train_x,train_y) , (test_x,test_y)= load_data()
    train_x = train_x.reshape(-1,28,28,1)
    test_x = test_x.reshape(-1,28,28,1)
    train_y = to_categorical(train_y,10)
    test_y = to_categorical(test_y,10)
    '''卷积神经网络部分'''
    model = Sequential()
    # model.add(Dense(input_dim=28*28,units=500,activation='sigmoid'))
    model.add(Conv2D(filters=2,kernel_size=(3,3),input_shape=(28,28,1),activation='relu'))
    # model.add(Dense(units=500,activation='sigmoid'))
    model.add(MaxPooling2D(pool_size=2, strides=2))
    model.add(Flatten())
    model.add(Dense(units=10,activation='softmax'))
    model.compile(optimizer='adam',loss='categorical_crossentropy',metrics=['accuracy'])
    model.fit(train_x,train_y,batch_size=100,epochs=20)
    '''效果评估部分'''
    result = model.evaluate(test_x,test_y)
    print('ACC:',result[1])
    # model.save('hw.h5')
    # model = load_model('hw.h5')
    '''胡乱测试部分，用画板画了一张图，看能否识别'''
    img = np.array(Image.open('test.png'))
    img *= 255
    img = img.reshape(1,28,28,1)
    print(img.shape)
    res = np.argmax(model.predict(img))
    print('res = ' , res)

根据训练好的模型可以识别出我手写的28*28像素的数字的为3：

运行结果为：