deep_neural_networks.py<\/a> \uff0c\u5185\u5bb9\u53ef\u53c2\u8003\uff1a<\/p>\n\n\n\ndeep_neural_networks.py<\/p>\n\n\n\n
import numpy as np
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
\u200b
def add_layer(inputs, in_size, out_size, activation_function=None):
W = tf.Variable(tf.random_normal([in_size, out_size]))
b = tf.Variable(tf.zeros([1, out_size]) + 0.01)
\u200b
Z = tf.matmul(inputs, W) + b
if activation_function is None:
outputs = Z
else:
outputs = activation_function(Z)
\u200b
return outputs
\u200b
\u200b
if __name__ == \"__main__\":
\u200b
MNIST = input_data.read_data_sets(\"mnist\", one_hot=True)
\u200b
learning_rate = 0.01
batch_size = 128
n_epochs = 70
\u200b
X = tf.placeholder(tf.float32, [batch_size, 784])
Y = tf.placeholder(tf.float32, [batch_size, 10])
\u200b
layer_dims = [784, 500, 500, 10]
layer_count = len(layer_dims)-1
layer_iter = X
\u200b
for l in range(1, layer_count): # layer [1,layer_count-1] is hidden layer
layer_iter = add_layer(layer_iter, layer_dims[l-1], layer_dims[l], activation_function=tf.nn.relu)
prediction = add_layer(layer_iter, layer_dims[layer_count-1], layer_dims[layer_count], activation_function=None)
\u200b
entropy = tf.nn.softmax_cross_entropy_with_logits(labels=Y, logits=prediction)
loss = tf.reduce_mean(entropy)
\u200b
optimizer = tf.train.GradientDescentOptimizer(learning_rate).minimize(loss)
\u200b
init = tf.initialize_all_variables()
\u200b
with tf.Session() as sess:
sess.run(init)
\u200b
n_batches = int(MNIST.test.num_examples\/batch_size)
for i in range(n_epochs):
for j in range(n_batches):
X_batch, Y_batch = MNIST.train.next_batch(batch_size)
_, loss_ = sess.run([optimizer, loss], feed_dict={X: X_batch, Y: Y_batch})
if i % 10 == 5 and j == 0:
print \"Loss of epochs[{0}]: {1}\".format(i, loss_)
\u200b
# test the model
n_batches = int(MNIST.test.num_examples\/batch_size)
total_correct_preds = 0
for i in range(n_batches):
X_batch, Y_batch = MNIST.test.next_batch(batch_size)
preds = sess.run(prediction, feed_dict={X: X_batch, Y: Y_batch})
correct_preds = tf.equal(tf.argmax(preds, 1), tf.argmax(Y_batch, 1))
accuracy = tf.reduce_sum(tf.cast(correct_preds, tf.float32))
\u200b
total_correct_preds += sess.run(accuracy)
\u200b
print \"Accuracy {0}\".format(total_correct_preds\/MNIST.test.num_examples)<\/pre>\n\n\n\n\u4ee3\u7801\u8bb2\u89e3<\/h4>\n\n\n\nadd_layer \u51fd\u6570<\/h4>\n\n\n\n
\u5141\u8bb8\u7528\u6237\u6307\u5b9a\u4e0a\u4e00\u5c42\u7684\u8f93\u51fa\u8282\u70b9\u7684\u4e2a\u6570\u4f5c\u4e3a input_size<\/strong>\uff0c\u672c\u5c42\u7684\u8282\u70b9\u4e2a\u6570\u4f5c\u4e3a output_size<\/strong>\uff0c\u5e76\u6307\u5b9a\u6fc0\u6d3b\u51fd\u6570 activation_function<\/strong> \u53ef\u4ee5\u770b\u5230\u6211\u4eec\u8c03\u7528\u7684\u65f6\u5019\u4f4d\u795e\u7ecf\u7f51\u7edc\u6dfb\u52a0\u4e86\u4e24\u4e2a \u9690\u85cf\u5c42<\/strong> \u548c \u8f93\u51fa\u5c42<\/strong><\/p>\n\n\n\nfor l in range(1, layer_count): # layer [1,layer_count-1] is hidden layer
layer_iter = add_layer(layer_iter, layer_dims[l-1], layer_dims[l], activation_function=tf.nn.relu)
prediction = add_layer(layer_iter, layer_dims[layer_count-1], layer_dims[layer_count], activation_function=None)
\u200b
entropy = tf.nn.softmax_cross_entropy_with_logits(labels=Y, logits=prediction)
loss = tf.reduce_mean(entropy)<\/pre>\n\n\n\n\u795e\u7ecf\u7f51\u7edc\u914d\u7f6e<\/h4>\n\n\n\n
\u6ce8\u610f\u8fd9\u4e00\u884c\uff0c\u6211\u4eec\u914d\u7f6e\u4e86\u4e00\u4e2a\u6df1\u5ea6\u7684\u795e\u7ecf\u7f51\u7edc\uff0c\u5b83\u5305\u542b\u4e24\u4e2a<\/strong>\u9690\u85cf\u5c42\uff0c\u4e00\u4e2a\u8f93\u5165\u5c42\u548c\u4e00\u4e2a\u8f93\u51fa\u5c42 \u9690\u85cf\u5c42\u7684\u8282\u70b9\u6570\u4e3a 500<\/strong><\/p>\n\n\n\nlayer_dims = [784, 500, 500, 10]<\/pre>\n\n\n\n
![\"image\"\/](\"https:\/\/share-10039692.file.myqcloud.com\/lab\/a17bac8dbb\/image\/jsaj98sn35\/neural_network.png\")
<\/figure>\n\n\n\n