Tensorflow学习总结（1）：CNN

    
    

# create pooling ,in order to reduce the loss of info when cutting the image
def max_pool_2x2(x):
    return tf.nn.max_pool(x, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')

tf.nn.max_pool(value, ksize, strides, padding, name=None)

第一个参数value：需要池化的输入，一般池化层接在卷积层后面，所以输入通常是feature map，依然是[batch, height, width, channels]这样的shape

第二个参数ksize：池化窗口的大小，取一个四维向量，一般是[1, height, width, 1]，因为我们不想在batch和channels上做池化，所以这两个维度设为了1

第三个参数strides：和卷积类似，窗口在每一个维度上滑动的步长，一般也是[1, stride,stride, 1]

第四个参数padding：和卷积类似，可以取'VALID' 或者'SAME'

返回一个Tensor，类型不变，shape仍然是[batch, height, width, channels]这种形式

    
    

import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data

# 60000行的训练数据集（mnist.train）和10000行的测试数据集（mnist.test）
# (每一行包含28*28=784个像素点）


# Import data
mnist = input_data.read_data_sets("MNIST_data/", one_hot=True)


# init weight
def weight_variable(shape):
    initial = tf.truncated_normal(shape, stddev=0.01)
    return tf.Variable(initial)


# init bias
def bias_variable(shape):
    initial = tf.constant(0.1, shape=shape)
    return tf.Variable(initial)


# create CNN layer
def conv2d(x, W):
    # stride [1,x_movement,y_movement,1],stride[0] and stride[3] must be 1
    return tf.nn.conv2d(x, W, strides=[1, 1, 1, 1], padding='SAME')                ### ???


# create pooling ,in order to reduce the loss of info when cutting the image
def max_pool_2x2(x):
    return tf.nn.max_pool(x, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')


# def compute_accuracy
def comput_accuracy(v_xs, v_ys):
    global prediction
    y_pre = sess.run(prediction, feed_dict={xs: v_xs, keep_prob: 1})
    correct_pre = tf.equal(tf.argmax(y_pre, 1), tf.argmax(v_ys, 1))
    accuracy = tf.reduce_mean(tf.cast(correct_pre, tf.float32))
    result = sess.run(accuracy, feed_dict={xs: v_xs, ys: v_ys, keep_prob: 1})
    return result


# define placeholder for inputs to network
xs = tf.placeholder(tf.float32, [None, 784])            # 28x28
ys = tf.placeholder(tf.float32, [None, 10])
keep_prob = tf.placeholder(tf.float32)
x_image = tf.reshape(xs, [-1, 28, 28, 1])
# print(x_image.shape)   #[n_sample.28,28,1]

# conv1 layer
W_conv1 = weight_variable([5, 5, 1, 32])        # patch 5x5,in size 1,out size 32
b_conv1 = bias_variable([32])
hide_conv1 = tf.nn.relu(conv2d(x_image, W_conv1) + b_conv1)
# relu: let data nonlinear , output size 28x28x32
hide_pool1 = max_pool_2x2(hide_conv1)                                   # output size 14x14x32

# conv2 layer
W_conv2 = weight_variable([5, 5, 32, 64])        # patch 5x5,in size 32,out size 64
b_conv2 = bias_variable([64])
hide_conv2 = tf.nn.relu(conv2d(hide_pool1, W_conv2) + b_conv2)            # relu: let data nonlinear , output size 14x14x64
hide_pool2 = max_pool_2x2(hide_conv2)                                   # output size 7x7x64

# func1 layer
W_fc1 = weight_variable([7*7*64, 1024])                   # 全连接
b_fc1 = bias_variable([1024])
# [n_samples,7,7,64] ->> [n_samples,7*7*64]
h_pool2_flat = tf.reshape(hide_pool2, [-1, 7*7*64])            # 转化为1维
h_fc1 = tf.nn.relu(tf.matmul(h_pool2_flat, W_fc1) + b_fc1)       # 点积
h_fc1_drop = tf.nn.dropout(h_fc1, keep_prob)

# func2 layer
W_fc2 = weight_variable([1024, 10])
b_fc2 = bias_variable([10])
prediction = tf.nn.softmax(tf.matmul(h_fc1_drop, W_fc2) + b_fc2)

# the error between the prediction and real data
cross_entropy = tf.reduce_mean(-tf.reduce_sum(ys*tf.log(prediction), reduction_indices=[1]))
train_step = tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)

# create session
with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    for i in range(1000):
        batch_xs, batch_ys = mnist.train.next_batch(100)
        sess.run(train_step, feed_dict={xs: batch_xs, ys: batch_ys, keep_prob: 0.5})
        if i % 40 == 0:
            print(comput_accuracy(mnist.test.images, mnist.test.labels))