'''CNN卷积神经网络'''
'''整个网络结构就是输入---卷积---池化---卷积---池化---全连接---输出'''
'''
1、convolutional layer1 + max pooling;
2、convolutional layer2 + max pooling;
3、fully connected layer1 + dropout;
4、fully connected layer2 to prediction.
'''
'''
1、一个卷积核对应一层,多个卷积核生成多层
2、一个卷积核(filter/patch/kernal)是:BatchSize * Height * Width * InputDepth * OutputDepth,层数的改变是通过OutputDepth来改变的
3、卷积过程需要用到卷积核,可以理解为一个二维的滑动窗口,每个卷积核由n * m个小格组成,每个小格都有自己的权重值,卷积过程就是将
卷积核覆盖的像素值乘以对应格子内的系数求和得到一个新的值,由于将n * m个点压缩成一个点,所以长宽会变小,每个卷积核扫描完后产
生一层图像,多个卷积核就产生多层,高度就会增加(也可以理解为不同卷积核/滤波器处理同一张图片后产生的多张图)。
4、stride步长,跨度为多少
-"VALID":输入大小不一样就舍去,输出比原图小,让卷积核使用保守原则,超过卷积核框架的边缘就丢弃
5、padding-|
-"SAME":输入大小不一样的时候要补0,输出和原图一样大
6、Pooling:相当于一个固化的卷积核,作用不是降维,而是缩小数据点,为了得到主要信息,忽略次要信息,
池化后长宽不变,高度(filter的个数)不变,max pooling 强调特征,
average pooling 强调背景
Alphago用的是补0
7、实际上是卷积窗口大小对应feature map,含义:[batch, height, weight, channels],因为在卷积池化过程中不需要对batch核channels操作,
所以第一个和最后一个是1
'''
'''CNN 代码'''
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
mnist = input_data.read_data_sets('MNIST_data', one_hot=True)
def compute_accuracy(v_xs, v_ys):
global prediction
y_pre = sess.run(prediction, feed_dict={xs: v_xs, keep_prob: 1})
correct_prediction = tf.equal(tf.argmax(y_pre, 1), tf.argmax(v_ys, 1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
result = sess.run(accuracy, feed_dict={xs: v_xs, ys: v_ys, keep_prob: 1})
return result
def weight_variable(shape):
initial = tf.truncated_normal(shape, stddev=0.1)
return tf.Variable(initial)
def bias_variable(shape):
initial = tf.constant(0.1, shape=shape)
return tf.Variable(initial)
def conv2d(x, W):
return tf.nn.conv2d(x, W, strides=[1, 1, 1, 1], padding='SAME')
def max_pool_2x2(x):
return tf.nn.max_pool(x, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
xs = tf.placeholder(tf.float32, [None, 784])/255.
ys = tf.placeholder(tf.float32, [None, 10])
keep_prob = tf.placeholder(tf.float32)
x_image = tf.reshape(xs, [-1, 28, 28, 1])
W_conv1 = weight_variable([5, 5, 1, 32])
b_conv1 = bias_variable([32])
h_conv1 = tf.nn.relu(conv2d(x_image, W_conv1) + b_conv1)
h_pool1 = max_pool_2x2(h_conv1)
W_conv2 = weight_variable([5, 5, 32, 64])
b_conv2 = bias_variable([64])
h_conv2 = tf.nn.relu(conv2d(h_pool1, W_conv2) + b_conv2)
h_pool2 = max_pool_2x2(h_conv2)
W_fc1 = weight_variable([7*7*64, 1024])
b_fc1 = bias_variable([1024])
h_pool2_flat = tf.reshape(h_pool2, [-1, 7*7*64])
h_fc1 = tf.nn.relu(tf.matmul(h_pool2_flat, W_fc1) + b_fc1)
h_fc1_drop = tf.nn.dropout(h_fc1, keep_prob)
W_fc2 = weight_variable([1024, 10])
b_fc2 = bias_variable([10])
prediction = tf.nn.softmax(tf.matmul(h_fc1_drop, W_fc2) + b_fc2)
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)
sess = tf.Session()
sess.run(tf.initialize_all_variables())
for i in range(1000):
batch_xs, batch_ys = mnist.train.next_batch(10)
sess.run(train_step, feed_dict={xs: batch_xs, ys: batch_ys, keep_prob: 0.5})
if i % 50 == 0:
print(compute_accuracy(
mnist.test.images, mnist.test.labels))
'''
运行结果:
0.105
0.5958
0.7556
0.8036
0.8297
0.8698
0.8807
0.8979
0.9058
0.9017
0.9179
0.9251
0.9261
0.9259
0.9272
0.9387
0.9382
0.9351
0.94
0.9434
'''