利用keras搭建神经卷积网络（CNN）

CNN

因为之前上课的时候对CNN学的不是很好，所以在这次训练之前我在知乎上找到一篇对CNN讲解的文章，先进行阅读了一番。来自机器之心的一篇文章

http://mp.weixin.qq.com/s?__biz=MzA3MzI4MjgzMw==&mid=2650717691&idx=2&sn=3f0b66aa9706aae1a30b01309aa0214c#rd

https://www.zhihu.com/question/52668301

阅读文章后，对代码理解就很easy了，我没有用莫烦教程里的代码，因为那个准确率没有官方文档给的mnist_cnn的代码高，贴一下官方代码地址https://github.com/fchollet/keras/blob/master/examples/mnist_cnn.py

#!/usr/bin/python# -*- coding:utf8 -*-'''Trains a simple convnet on the MNIST dataset.Gets to 99.25% test accuracy after 12 epochs(there is still a lot of margin for parameter tuning).16 seconds per epoch on a GRID K520 GPU.'''from __future__ import print_functionimport kerasfrom keras.datasets import mnistfrom keras.models import Sequentialfrom keras.layers import Dense, Dropout, Flatten, Conv2D, MaxPooling2Dfrom keras import backend as Kbatch_size = 128num_classes = 10epochs = 12# pretreatment 预处理# input image dimensionsimg_rows, img_cols = 28, 28# the data, shuffled and split between train and test sets(x_train, y_train), (x_test, y_test) = mnist.load_data()if K.image_data_format() == 'channels_first': #判断图片格式是 channel在前还是在后（channel：黑白为1,彩色为3）    x_train = x_train.reshape(x_train.shape[0], 1, img_rows, img_cols) #shape[0]指例子的个数    x_test = x_test.reshape(x_test.shape[0], 1, img_rows, img_cols)    input_shape = (1, img_rows, img_cols)else:    x_train = x_train.reshape(x_train.shape[0], img_rows, img_cols, 1)    x_test = x_test.reshape(x_test.shape[0], img_rows, img_cols, 1)    input_shape = (img_rows, img_cols, 1)x_train = x_train.astype('float32')x_test = x_test.astype('float32')x_train /= 255x_test /= 255print('x_train shape:', x_train.shape)print(x_train.shape[0], 'train samples')print(x_test.shape[0], 'test samples')# convert class vectors to binary class matricesy_train = keras.utils.to_categorical(y_train, num_classes)y_test = keras.utils.to_categorical(y_test, num_classes)# build the neural net 建模型(卷积—relu-卷积-relu-池化-relu-卷积-relu-池化-全连接)model = Sequential()model.add(Conv2D(32, kernel_size=(3, 3),                 activation='relu',                 input_shape=input_shape)) # 32个过滤器，过滤器大小是3×3，32×26×26model.add(Conv2D(64, (3, 3), activation='relu')) #64×24×24model.add(MaxPooling2D(pool_size=(2, 2)))# 向下取样model.add(Dropout(0.25))model.add(Flatten()) #降维：将64×12×12降为1维（即把他们相乘起来）model.add(Dense(128, activation='relu'))model.add(Dropout(0.5))model.add(Dense(num_classes, activation='softmax')) #全连接2层# compile the model 编译模型model.compile(loss=keras.losses.categorical_crossentropy,              optimizer=keras.optimizers.Adadelta(),              metrics=['accuracy'])# train the model 训练模型model.fit(x_train, y_train,          batch_size=batch_size,          epochs=epochs,          verbose=1,          validation_data=(x_test, y_test))# test the model 测试模型score = model.evaluate(x_test, y_test, verbose=0)print('Test loss:', score[0])print('Test accuracy:', score[1])

训练结果
经过12个epoch后