輸入量與輸出量都是數(shù)據(jù)原有的全部特征，我們利用tensorflow的optimizer對(duì)w1ij和w2ji進(jìn)行優(yōu)化。在優(yōu)化的最后，w1ij就是我們將n維數(shù)據(jù)編碼到2維的編碼方式，w2ji就是我們將2維數(shù)據(jù)進(jìn)行解碼到n維數(shù)據(jù)的解碼方式。

3、python中encode的實(shí)現(xiàn)

def encoder(x):
    layer_1 = tf.nn.sigmoid(tf.add(tf.matmul(x, weights['encoder_h1']),
                                   biases['encoder_b1']))
    layer_2 = tf.nn.sigmoid(tf.add(tf.matmul(layer_1, weights['encoder_h2']),
                                   biases['encoder_b2']))
    layer_3 = tf.nn.sigmoid(tf.add(tf.matmul(layer_2, weights['encoder_h3']),
                                   biases['encoder_b3']))
    layer_4 = tf.add(tf.matmul(layer_3, weights['encoder_h4']),
                                    biases['encoder_b4'])
    return layer_4
def decoder(x):
    layer_1 = tf.nn.sigmoid(tf.add(tf.matmul(x, weights['decoder_h1']),
                                   biases['decoder_b1']))
    layer_2 = tf.nn.sigmoid(tf.add(tf.matmul(layer_1, weights['decoder_h2']),
                                   biases['decoder_b2']))
    layer_3 = tf.nn.sigmoid(tf.add(tf.matmul(layer_2, weights['decoder_h3']),
                                biases['decoder_b3']))
    layer_4 = tf.nn.sigmoid(tf.add(tf.matmul(layer_3, weights['decoder_h4']),
                                biases['decoder_b4']))
    return layer_4
encoder_op = encoder(X)
decoder_op = decoder(encoder_op)

其中encode函數(shù)的輸出就是編碼后的結(jié)果。

全部代碼

該例子為手寫體識(shí)別例子，將784維縮小為2維，并且以圖像的方式顯示。

import matplotlib.pyplot as plt
import numpy as np
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
mnist = input_data.read_data_sets("MNIST_data",one_hot = "true")
learning_rate = 0.01    #學(xué)習(xí)率
training_epochs = 10	#訓(xùn)練十次
batch_size = 256
display_step = 1
examples_to_show = 10
n_input = 784
X = tf.placeholder(tf.float32,[None,n_input])
#encode的過程分為4次，分別是784->128、128->64、64->10、10->2
n_hidden_1 = 128
n_hidden_2 = 64
n_hidden_3 = 10
n_hidden_4 = 2
weights = {
    #這四個(gè)是用于encode的
    'encoder_h1': tf.Variable(tf.truncated_normal([n_input, n_hidden_1],)),
    'encoder_h2': tf.Variable(tf.truncated_normal([n_hidden_1, n_hidden_2],)),
    'encoder_h3': tf.Variable(tf.truncated_normal([n_hidden_2, n_hidden_3],)),
    'encoder_h4': tf.Variable(tf.truncated_normal([n_hidden_3, n_hidden_4],)),
    #這四個(gè)是用于decode的
    'decoder_h1': tf.Variable(tf.truncated_normal([n_hidden_4, n_hidden_3],)),
    'decoder_h2': tf.Variable(tf.truncated_normal([n_hidden_3, n_hidden_2],)),
    'decoder_h3': tf.Variable(tf.truncated_normal([n_hidden_2, n_hidden_1],)),
    'decoder_h4': tf.Variable(tf.truncated_normal([n_hidden_1, n_input],)),
}
biases = {
    #這四個(gè)是用于encode的
    'encoder_b1': tf.Variable(tf.random_normal([n_hidden_1])),
    'encoder_b2': tf.Variable(tf.random_normal([n_hidden_2])),
    'encoder_b3': tf.Variable(tf.random_normal([n_hidden_3])),
    'encoder_b4': tf.Variable(tf.random_normal([n_hidden_4])),
    #這四個(gè)是用于decode的
    'decoder_b1': tf.Variable(tf.random_normal([n_hidden_3])),
    'decoder_b2': tf.Variable(tf.random_normal([n_hidden_2])),
    'decoder_b3': tf.Variable(tf.random_normal([n_hidden_1])),
    'decoder_b4': tf.Variable(tf.random_normal([n_input])),
}
def encoder(x):
    #encode函數(shù)，分為四步，layer4為編碼后的結(jié)果
    layer_1 = tf.nn.sigmoid(tf.add(tf.matmul(x, weights['encoder_h1']),
                                   biases['encoder_b1']))
    layer_2 = tf.nn.sigmoid(tf.add(tf.matmul(layer_1, weights['encoder_h2']),
                                   biases['encoder_b2']))
    layer_3 = tf.nn.sigmoid(tf.add(tf.matmul(layer_2, weights['encoder_h3']),
                                   biases['encoder_b3']))
    layer_4 = tf.add(tf.matmul(layer_3, weights['encoder_h4']),
                                    biases['encoder_b4'])
    return layer_4
def decoder(x):
    #decode函數(shù)，分為四步，layer4為解碼后的結(jié)果
    layer_1 = tf.nn.sigmoid(tf.add(tf.matmul(x, weights['decoder_h1']),
                                   biases['decoder_b1']))
    layer_2 = tf.nn.sigmoid(tf.add(tf.matmul(layer_1, weights['decoder_h2']),
                                   biases['decoder_b2']))
    layer_3 = tf.nn.sigmoid(tf.add(tf.matmul(layer_2, weights['decoder_h3']),
                                biases['decoder_b3']))
    layer_4 = tf.nn.sigmoid(tf.add(tf.matmul(layer_3, weights['decoder_h4']),
                                biases['decoder_b4']))
    return layer_4
encoder_op = encoder(X)
decoder_op = decoder(encoder_op)
#將編碼再解碼的結(jié)果與原始碼對(duì)比，查看區(qū)別
y_pred = decoder_op
y_label = X
#比較特征損失情況
cost = tf.reduce_mean(tf.square(y_pred-y_label))
train = tf.train.AdamOptimizer(learning_rate).minimize(cost)
init = tf.global_variables_initializer()
with tf.Session() as sess:
    sess.run(init)
    #每個(gè)世代進(jìn)行total_batch次訓(xùn)練
    total_batch = int(mnist.train.num_examples/batch_size)
    for epoch in range(training_epochs):
        for i in range(total_batch):
            batch_xs,batch_ys = mnist.train.next_batch(batch_size)
            _,c = sess.run([train,cost],feed_dict={X:batch_xs})
        if epoch % display_step == 0:
            print("Epoch :","%02d"%epoch,"cost =","%.4f"%c)
    #利用test測(cè)試機(jī)進(jìn)行測(cè)試
    encoder_result = sess.run(encoder_op,feed_dict={X:mnist.test.images})
    plt.scatter(encoder_result[:,0],encoder_result[:,1],c=np.argmax(mnist.test.labels,1),s=1)
    plt.show()

實(shí)現(xiàn)結(jié)果為：

可以看到實(shí)驗(yàn)結(jié)果分為很多個(gè)區(qū)域塊，基本可以識(shí)別。

以上就是python神經(jīng)網(wǎng)絡(luò)使用tensorflow實(shí)現(xiàn)自編碼Autoencoder的詳細(xì)內(nèi)容，更多關(guān)于tensorflow自編碼Autoencoder的資料請(qǐng)關(guān)注腳本之家其它相關(guān)文章！

您可能感興趣的文章:

亚洲乱码中文字幕综合,中国熟女仑乱hd,亚洲精品乱拍国产一区二区三区,一本大道卡一卡二卡三乱码全集资源,又粗又黄又硬又爽的免费视频

軟件下載

源碼下載

軟件編程

網(wǎng)絡(luò)編程

在線工具

數(shù)據(jù)庫

CMS

常用工具

python神經(jīng)網(wǎng)絡(luò)使用tensorflow實(shí)現(xiàn)自編碼Autoencoder

目錄

學(xué)習(xí)前言