吴裕雄--天生自然TensorFlow高层封装：使用TFLearn处理MNIST数据集实现LeNet-5模型

# 1. 通过TFLearn的API定义卷机神经网络。

import tflearn
import tflearn.datasets.mnist as mnist

from tflearn.layers.conv import conv_2d, max_pool_2d
from tflearn.layers.estimator import regression
from tflearn.layers.core import input_data, dropout, fully_connected

trainX, trainY, testX, testY = mnist.load_data(data_dir="F:\\TensorFlowGoogle\\201806-github\\datasets\\MNIST_data", one_hot=True)
# 将图像数据resize成卷积卷积神经网络输入的格式。
trainX = trainX.reshape([-1, 28, 28, 1])
testX = testX.reshape([-1, 28, 28, 1])

# 构建神经网络。
net = input_data(shape=[None, 28, 28, 1], name='input')
net = conv_2d(net, 32, 5, activation='relu')
net = max_pool_2d(net, 2)
net = conv_2d(net, 64, 5, activation='relu')
net = max_pool_2d(net, 2)
net = fully_connected(net, 500, activation='relu')
net = fully_connected(net, 10, activation='softmax')
# 定义学习任务。指定优化器为sgd，学习率为0.01，损失函数为交叉熵。
net = regression(net, optimizer='sgd', learning_rate=0.01,loss='categorical_crossentropy')

# 2. 通过TFLearn的API训练神经网络。
# 通过定义的网络结构训练模型，并在指定的验证数据上验证模型的效果。
model = tflearn.DNN(net, tensorboard_verbose=0)
model.fit(trainX, trainY, n_epoch=10,validation_set=([testX, testY]),show_metric=True)