tensorflow tutorials:用tensorflow实现卷积神经网络(Convolutional Neural Networks)

Posted 风雪夜归子

tags:

篇首语:本文由小常识网(cha138.com)小编为大家整理,主要介绍了tensorflow tutorials:用tensorflow实现卷积神经网络(Convolutional Neural Networks)相关的知识,希望对你有一定的参考价值。


声明:版权所有,转载请联系作者并注明出处:  http://blog.csdn.net/u013719780?viewmode=contents



from __future__ import print_function

import tensorflow as tf

# Import MNIST data
from tensorflow.examples.tutorials.mnist import input_data
mnist = input_data.read_data_sets("/tmp/data/", one_hot=True)

# Parameters
learning_rate = 0.001
training_iters = 200000
batch_size = 128
display_step = 10

# Network Parameters
n_input = 784 # MNIST data input (img shape: 28*28)
n_classes = 10 # MNIST total classes (0-9 digits)
dropout = 0.75 # Dropout, probability to keep units

# tf Graph input
x = tf.placeholder(tf.float32, [None, n_input])
y = tf.placeholder(tf.float32, [None, n_classes])
keep_prob = tf.placeholder(tf.float32) #dropout (keep probability)


# Create some wrappers for simplicity
def conv2d(x, W, b, strides=1):
    # Conv2D wrapper, with bias and relu activation
    x = tf.nn.conv2d(x, W, strides=[1, strides, strides, 1], padding='SAME')
    x = tf.nn.bias_add(x, b)
    return tf.nn.relu(x)


def maxpool2d(x, k=2):
    # MaxPool2D wrapper
    return tf.nn.max_pool(x, ksize=[1, k, k, 1], strides=[1, k, k, 1],
                          padding='SAME')


# Create model
def conv_net(x, weights, biases, dropout):
    # Reshape input picture
    x = tf.reshape(x, shape=[-1, 28, 28, 1])

    # Convolution Layer
    conv1 = conv2d(x, weights['wc1'], biases['bc1'])
    # Max Pooling (down-sampling)
    conv1 = maxpool2d(conv1, k=2)

    # Convolution Layer
    conv2 = conv2d(conv1, weights['wc2'], biases['bc2'])
    # Max Pooling (down-sampling)
    conv2 = maxpool2d(conv2, k=2)

    # Fully connected layer
    # Reshape conv2 output to fit fully connected layer input
    fc1 = tf.reshape(conv2, [-1, weights['wd1'].get_shape().as_list()[0]])
    fc1 = tf.add(tf.matmul(fc1, weights['wd1']), biases['bd1'])
    fc1 = tf.nn.relu(fc1)
    # Apply Dropout
    fc1 = tf.nn.dropout(fc1, dropout)

    # Output, class prediction
    out = tf.add(tf.matmul(fc1, weights['out']), biases['out'])
    return out

# Store layers weight & bias
weights = 
    # 5x5 conv, 1 input, 32 outputs
    'wc1': tf.Variable(tf.random_normal([5, 5, 1, 32])),
    # 5x5 conv, 32 inputs, 64 outputs
    'wc2': tf.Variable(tf.random_normal([5, 5, 32, 64])),
    # fully connected, 7*7*64 inputs, 1024 outputs
    'wd1': tf.Variable(tf.random_normal([7*7*64, 1024])),
    # 1024 inputs, 10 outputs (class prediction)
    'out': tf.Variable(tf.random_normal([1024, n_classes]))


biases = 
    'bc1': tf.Variable(tf.random_normal([32])),
    'bc2': tf.Variable(tf.random_normal([64])),
    'bd1': tf.Variable(tf.random_normal([1024])),
    'out': tf.Variable(tf.random_normal([n_classes]))


# Construct model
pred = conv_net(x, weights, biases, keep_prob)

# Define loss and optimizer
cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(pred, y))
optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate).minimize(cost)

# Evaluate model
correct_pred = tf.equal(tf.argmax(pred, 1), tf.argmax(y, 1))
accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))

# Initializing the variables
init = tf.initialize_all_variables()

# Launch the graph
with tf.Session() as sess:
    sess.run(init)
    step = 1
    # Keep training until reach max iterations
    while step * batch_size < training_iters:
        batch_x, batch_y = mnist.train.next_batch(batch_size)
        # Run optimization op (backprop)
        sess.run(optimizer, feed_dict=x: batch_x, y: batch_y,
                                       keep_prob: dropout)
        if step % display_step == 0:
            # Calculate batch loss and accuracy
            loss, acc = sess.run([cost, accuracy], feed_dict=x: batch_x,
                                                              y: batch_y,
                                                              keep_prob: 1.)
            print("Iter " + str(step*batch_size) + ", Minibatch Loss= " + \\
                  ":.6f".format(loss) + ", Training Accuracy= " + \\
                  ":.5f".format(acc))
        step += 1
    print("Optimization Finished!")

    # Calculate accuracy for 256 mnist test images
    print("Testing Accuracy:", \\
        sess.run(accuracy, feed_dict=x: mnist.test.images[:256],
                                      y: mnist.test.labels[:256],
                                      keep_prob: 1.))
Extracting /tmp/data/train-images-idx3-ubyte.gz
Extracting /tmp/data/train-labels-idx1-ubyte.gz
Extracting /tmp/data/t10k-images-idx3-ubyte.gz
Extracting /tmp/data/t10k-labels-idx1-ubyte.gz
Iter 1280, Minibatch Loss= 33119.816406, Training Accuracy= 0.21875
Iter 2560, Minibatch Loss= 16051.991211, Training Accuracy= 0.36719
Iter 3840, Minibatch Loss= 10761.388672, Training Accuracy= 0.60938
Iter 5120, Minibatch Loss= 4721.864746, Training Accuracy= 0.75781
Iter 6400, Minibatch Loss= 2037.779785, Training Accuracy= 0.85156
Iter 7680, Minibatch Loss= 7413.786133, Training Accuracy= 0.74219
Iter 8960, Minibatch Loss= 3249.669434, Training Accuracy= 0.83594
Iter 10240, Minibatch Loss= 4321.263672, Training Accuracy= 0.81250
Iter 11520, Minibatch Loss= 1479.329956, Training Accuracy= 0.92188
Iter 12800, Minibatch Loss= 2355.179688, Training Accuracy= 0.84375
Iter 14080, Minibatch Loss= 1105.266357, Training Accuracy= 0.92188
Iter 15360, Minibatch Loss= 1615.281494, Training Accuracy= 0.89844
Iter 16640, Minibatch Loss= 2617.854248, Training Accuracy= 0.89062
Iter 17920, Minibatch Loss= 1187.729980, Training Accuracy= 0.92188
Iter 19200, Minibatch Loss= 919.462341, Training Accuracy= 0.89844
Iter 20480, Minibatch Loss= 257.105347, Training Accuracy= 0.96094
Iter 21760, Minibatch Loss= 2607.273438, Training Accuracy= 0.85156
Iter 23040, Minibatch Loss= 793.234375, Training Accuracy= 0.95312
Iter 24320, Minibatch Loss= 1252.133911, Training Accuracy= 0.91406
Iter 25600, Minibatch Loss= 1356.419678, Training Accuracy= 0.91406
Iter 26880, Minibatch Loss= 894.269165, Training Accuracy= 0.95312
Iter 28160, Minibatch Loss= 1081.752197, Training Accuracy= 0.89062
Iter 29440, Minibatch Loss= 1214.221924, Training Accuracy= 0.92969
Iter 30720, Minibatch Loss= 1026.619263, Training Accuracy= 0.90625
Iter 32000, Minibatch Loss= 895.326416, Training Accuracy= 0.93750
Iter 33280, Minibatch Loss= 996.706665, Training Accuracy= 0.93750
Iter 34560, Minibatch Loss= 1258.645020, Training Accuracy= 0.92969
Iter 35840, Minibatch Loss= 757.918152, Training Accuracy= 0.95312
Iter 37120, Minibatch Loss= 1724.612305, Training Accuracy= 0.90625
Iter 38400, Minibatch Loss= 63.499573, Training Accuracy= 0.99219
Iter 39680, Minibatch Loss= 180.159348, Training Accuracy= 0.97656
Iter 40960, Minibatch Loss= 1820.881470, Training Accuracy= 0.90625
Iter 42240, Minibatch Loss= 740.729248, Training Accuracy= 0.96094
Iter 43520, Minibatch Loss= 452.834045, Training Accuracy= 0.95312
Iter 44800, Minibatch Loss= 168.801666, Training Accuracy= 0.97656
Iter 46080, Minibatch Loss= 149.422440, Training Accuracy= 0.98438
Iter 47360, Minibatch Loss= 1192.790039, Training Accuracy= 0.92969
Iter 48640, Minibatch Loss= 1069.188232, Training Accuracy= 0.93750
Iter 49920, Minibatch Loss= 578.359985, Training Accuracy= 0.92188
Iter 51200, Minibatch Loss= 185.997269, Training Accuracy= 0.96875
Iter 52480, Minibatch Loss= 269.750519, Training Accuracy= 0.97656
Iter 53760, Minibatch Loss= 188.991730, Training Accuracy= 0.97656
Iter 55040, Minibatch Loss= 1715.177734, Training Accuracy= 0.95312
Iter 56320, Minibatch Loss= 294.539795, Training Accuracy= 0.94531
Iter 57600, Minibatch Loss= 955.764709, Training Accuracy= 0.92969
Iter 58880, Minibatch Loss= 176.014526, Training Accuracy= 0.97656
Iter 60160, Minibatch Loss= 1202.663574, Training Accuracy= 0.94531
Iter 61440, Minibatch Loss= 1249.316406, Training Accuracy= 0.91406
Iter 62720, Minibatch Loss= 642.846130, Training Accuracy= 0.95312
Iter 64000, Minibatch Loss= 380.056335, Training Accuracy= 0.94531
Iter 65280, Minibatch Loss= 826.421265, Training Accuracy= 0.92969
Iter 66560, Minibatch Loss= 294.462433, Training Accuracy= 0.96875
Iter 67840, Minibatch Loss= 588.712280, Training Accuracy= 0.96875
Iter 69120, Minibatch Loss= 703.491882, Training Accuracy= 0.94531
Iter 70400, Minibatch Loss= 467.652283, Training Accuracy= 0.97656
Iter 71680, Minibatch Loss= 412.892883, Training Accuracy= 0.96875
Iter 72960, Minibatch Loss= 516.359497, Training Accuracy= 0.96875
Iter 74240, Minibatch Loss= 483.566406, Training Accuracy= 0.95312
Iter 75520, Minibatch Loss= 138.074493, Training Accuracy= 0.97656
Iter 76800, Minibatch Loss= 400.362335, Training Accuracy= 0.96875
Iter 78080, Minibatch Loss= 729.972046, Training Accuracy= 0.92188
Iter 79360, Minibatch Loss= 288.753296, Training Accuracy= 0.98438
Iter 80640, Minibatch Loss= 34.889206, Training Accuracy= 0.97656
Iter 81920, Minibatch Loss= 446.821838, Training Accuracy= 0.94531
Iter 83200, Minibatch Loss= 53.508987, Training Accuracy= 0.99219
Iter 84480, Minibatch Loss= 362.887177, Training Accuracy= 0.96094
Iter 85760, Minibatch Loss= 418.652954, Training Accuracy= 0.95312
Iter 87040, Minibatch Loss= 92.926300, Training Accuracy= 0.97656
Iter 88320, Minibatch Loss= 314.797424, Training Accuracy= 0.96875
Iter 89600, Minibatch Loss= 295.765839, Training Accuracy= 0.96875
Iter 90880, Minibatch Loss= 638.518188, Training Accuracy= 0.92969
Iter 92160, Minibatch Loss= 525.749329, Training Accuracy= 0.96094
Iter 93440, Minibatch Loss= 409.661530, Training Accuracy= 0.96094
Iter 94720, Minibatch Loss= 397.676514, Training Accuracy= 0.95312
Iter 96000, Minibatch Loss= 327.677795, Training Accuracy= 0.96094
Iter 97280, Minibatch Loss= 580.521729, Training Accuracy= 0.96875
Iter 98560, Minibatch Loss= 42.764221, Training Accuracy= 0.96875
Iter 99840, Minibatch Loss= 293.447510, Training Accuracy= 0.96094
Iter 101120, Minibatch Loss= 233.889969, Training Accuracy= 0.98438
Iter 102400, Minibatch Loss= 300.799316, Training Accuracy= 0.96094
Iter 103680, Minibatch Loss= 210.885757, Training Accuracy= 0.96094
Iter 104960, Minibatch Loss= 654.990173, Training Accuracy= 0.93750
Iter 106240, Minibatch Loss= 291.870728, Training Accuracy= 0.94531
Iter 107520, Minibatch Loss= 587.544617, Training Accuracy= 0.94531
Iter 108800, Minibatch Loss= 177.339050, Training Accuracy= 0.96094
Iter 110080, Minibatch Loss= 393.805206, Training Accuracy= 0.95312
Iter 111360, Minibatch Loss= 113.489090, Training Accuracy= 0.95312
Iter 112640, Minibatch Loss= 278.391144, Training Accuracy= 0.95312
Iter 113920, Minibatch Loss= 64.654800, Training Accuracy= 0.99219
Iter 115200, Minibatch Loss= 272.650635, Training Accuracy= 0.96875
Iter 116480, Minibatch Loss= 248.082993, Training Accuracy= 0.97656
Iter 117760, Minibatch Loss= 25.483871, Training Accuracy= 0.99219
Iter 119040, Minibatch Loss= 832.794373, Training Accuracy= 0.92188
Iter 120320, Minibatch Loss= 113.149963, Training Accuracy= 0.99219
Iter 121600, Minibatch Loss= 137.738678, Training Accuracy= 0.98438
Iter 122880, Minibatch Loss= 439.732605, Training Accuracy= 0.95312
Iter 124160, Minibatch Loss= 283.381012, Training Accuracy= 0.96875
Iter 125440, Minibatch Loss= 361.409546, Training Accuracy= 0.94531
Iter 126720, Minibatch Loss= 101.087547, Training Accuracy= 0.98438
Iter 128000, Minibatch Loss= 308.690063, Training Accuracy= 0.96094
Iter 129280, Minibatch Loss= 188.306870, Training Accuracy= 0.96094
Iter 130560, Minibatch Loss= 0.000000, Training Accuracy= 1.00000
Iter 131840, Minibatch Loss= 300.978424, Training Accuracy= 0.96875
Iter 133120, Minibatch Loss= 260.767548, Training Accuracy= 0.96094
Iter 134400, Minibatch Loss= 520.364746, Training Accuracy= 0.95312
Iter 135680, Minibatch Loss= 26.482552, Training Accuracy= 0.99219
Iter 136960, Minibatch Loss= 576.294434, Training Accuracy= 0.95312
Iter 138240, Minibatch Loss= 103.803009, Training Accuracy= 0.97656
Iter 139520, Minibatch Loss= 108.436554, Training Accuracy= 0.99219
Iter 140800, Minibatch Loss= 159.441193, Training Accuracy= 0.96875
Iter 142080, Minibatch Loss= 193.125519, Training Accuracy= 0.98438
Iter 143360, Minibatch Loss= 188.117294, Training Accuracy= 0.96875
Iter 144640, Minibatch Loss= 0.000000, Training Accuracy= 1.00000
Iter 145920, Minibatch Loss= 321.186157, Training Accuracy= 0.96094
Iter 147200, Minibatch Loss= 437.349396, Training Accuracy= 0.95312
Iter 148480, Minibatch Loss= 27.928650, Training Accuracy= 0.99219
Iter 149760, Minibatch Loss= 159.316650, Training Accuracy= 0.99219
Iter 151040, Minibatch Loss= 218.392944, Training Accuracy= 0.95312
Iter 152320, Minibatch Loss= 86.327057, Training Accuracy= 0.96094
Iter 153600, Minibatch Loss= 187.457947, Training Accuracy= 0.95312
Iter 154880, Minibatch Loss= 147.812164, Training Accuracy= 0.95312
Iter 156160, Minibatch Loss= 273.637848, Training Accuracy= 0.96094
Iter 157440, Minibatch Loss= 624.448669, Training Accuracy= 0.95312
Iter 158720, Minibatch Loss= 398.916992, Training Accuracy= 0.96094
Iter 160000, Minibatch Loss= 302.056213, Training Accuracy= 0.96875
Iter 161280, Minibatch Loss= 141.484192, Training Accuracy= 0.97656
Iter 162560, Minibatch Loss= 500.844543, Training Accuracy= 0.96875
Iter 163840, Minibatch Loss= 120.265915, Training Accuracy= 0.97656
Iter 165120, Minibatch Loss= 67.206924, Training Accuracy= 0.97656
Iter 166400, Minibatch Loss= 303.770020, Training Accuracy= 0.98438
Iter 167680, Minibatch Loss= 302.175598, Training Accuracy= 0.98438
Iter 168960, Minibatch Loss= 14.445847, Training Accuracy= 0.98438
Iter 170240, Minibatch Loss= 174.764893, Training Accuracy= 0.97656
Iter 171520, Minibatch Loss= 87.963837, Training Accuracy= 0.97656
Iter 172800, Minibatch Loss= 268.049377, Training Accuracy= 0.98438
Iter 174080, Minibatch Loss= 123.035660, Training Accuracy= 0.96875
Iter 175360, Minibatch Loss= 30.370827, Training Accuracy= 0.98438
Iter 176640, Minibatch Loss= 41.883797, Training Accuracy= 0.98438
Iter 177920, Minibatch Loss= 113.069115, Training Accuracy= 0.97656
Iter 179200, Minibatch Loss= 592.399658, Training Accuracy= 0.93750
Iter 180480, Minibatch Loss= 21.242783, Training Accuracy= 0.99219
Iter 181760, Minibatch Loss= 67.023407, Training Accuracy= 0.99219
Iter 183040, Minibatch Loss= 140.905319, Training Accuracy= 0.97656
Iter 184320, Minibatch Loss= 196.006165, Training Accuracy= 0.96875
Iter 185600, Minibatch Loss= 80.115158, Training Accuracy= 0.97656
Iter 186880, Minibatch Loss= 67.482613, Training Accuracy= 0.98438
Iter 188160, Minibatch Loss= 20.215111, Training Accuracy= 0.99219
Iter 189440, Minibatch Loss= 60.191788, Training Accuracy= 0.98438
Iter 190720, Minibatch Loss= 68.743011, Training Accuracy= 0.99219
Iter 192000, Minibatch Loss= 43.774590, Training Accuracy= 0.98438
Iter 193280, Minibatch Loss= 172.976425, Training Accuracy= 0.98438
Iter 194560, Minibatch Loss= 78.267181, Training Accuracy= 0.98438
Iter 195840, Minibatch Loss= 250.249496, Training Accuracy= 0.97656
Iter 197120, Minibatch Loss= 119.354599, Training Accuracy= 0.97656
Iter 198400, Minibatch Loss= 138.678864, Training Accuracy= 0.96875
Iter 199680, Minibatch Loss= 13.272423, Training Accuracy= 0.98438
Optimization Finished!
Testing Accuracy: 0.988281

以上是关于tensorflow tutorials:用tensorflow实现卷积神经网络(Convolutional Neural Networks)的主要内容,如果未能解决你的问题,请参考以下文章

tensorflow tutorials:用tensorflow实现卷积神经网络(Convolutional Neural Networks)

TensorFlow tutorial

tensorflow tutorials:用tensorflow实现降噪自编码器(Denoising Auto-Encoder)

TensorFlow Tutorial: 10 minutes Practical TensorFlow lesson for quick learners(译)

用 TensorFlow 压缩神经网络

tensorflow的安装问题 求解