Tensorflow 命名空间与计算图可视化

import tensorflow as tf
# 不同的命名空间
with tf.variable_scope("foo"):
    # 在命名空间foo下获取变量"bar",于是得到的变量名称为"foo/bar"
    a = tf.get_variable("bar", [1])
    print(a.name)
    # foo/bar:0
with tf.variable_scope("bar"):
    # 在命名空间bar下获取变量"bar",于是得到的变量名称为"bar/bar".此时变量在"bar/bar"和变量"foo/bar"并不冲突，于是可以正常运行
    b = tf.get_variable("bar", [1])
    print(b.name)
    # bar/bar:0
#  tf.Variable和tf.get_variable的区别。

with tf.name_scope("a"):
    # 使用tf.Variable函数生成变量时会受到tf.name_scope影响，于是这个变量的名称为"a/Variable"
    a = tf.Variable([1])
    print(a.name)
    # a/Variable: 0

    # tf.get_variable函数不受头tf.name_scope函数的影响，于是变量并不在a这个命名空间中
    a = tf.get_variable("b", [1])
    print(a.name)
    # b:0

# with tf.name_scope("b"):
    # 因为tf.get_variable不受tf.name_scope影响,所以这里将试图获取名称为"a"的变量。然而这个变量已经被声明了,于是这里会报重复声明的错误。
    # tf.get_variable("b",[1])

import tensorflow as tf
with tf.name_scope("input1"):
    input1 = tf.constant([1.0, 2.0, 3.0], name="input2")
with tf.name_scope("input2"):
    input2 = tf.Variable(tf.random_uniform([3]), name="input2")
output = tf.add_n([input1, input2], name="add")

writer = tf.summary.FileWriter("log/simple_example.log", tf.get_default_graph())
writer.close()

import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
# mnist_inference中定义的常量和前向传播的函数不需要改变，因为前向传播已经通过
# tf.variable_scope实现了计算节点按照网络结构的划分
import mnist_inference

# #### 1. 定义神经网络的参数。
BATCH_SIZE = 100
LEARNING_RATE_BASE = 0.8
LEARNING_RATE_DECAY = 0.99
REGULARIZATION_RATE = 0.0001
TRAINING_STEPS = 3000
MOVING_AVERAGE_DECAY = 0.99


# #### 2. 定义训练的过程并保存TensorBoard的log文件。
def train(mnist):
    #  将处理输入数据的计算都放在名字为"input"的命名空间中
    with tf.name_scope('input'):
        x = tf.placeholder(tf.float32, [None, mnist_inference.INPUT_NODE], name='x-input')
        y_ = tf.placeholder(tf.float32, [None, mnist_inference.OUTPUT_NODE], name='y-input')
    regularizer = tf.contrib.layers.l2_regularizer(REGULARIZATION_RATE)
    y = mnist_inference.inference(x, regularizer)
    global_step = tf.Variable(0, trainable=False)

    # 将处理滑动平均相关的计算都放在名为moving average 的命名空间下。
    with tf.name_scope("moving_average"):
        variable_averages = tf.train.ExponentialMovingAverage(MOVING_AVERAGE_DECAY, global_step)
        variables_averages_op = variable_averages.apply(tf.trainable_variables())  # 对可训练变量集合使用滑动平均

    # 将计算损失函数相关的计算都放在名为loss function 的命名空间下。
    with tf.name_scope("loss_function"):
        cross_entropy = tf.nn.sparse_softmax_cross_entropy_with_logits(logits=y, labels=tf.argmax(y_, 1))
        cross_entropy_mean = tf.reduce_mean(cross_entropy)
        # 将交叉熵加上权值的正则化
        loss = cross_entropy_mean + tf.add_n(tf.get_collection('losses'))

    # 将定义学习率、优化方法以及每一轮训练需要执行的操作都放在名字为"train_step"的命名空间下。
    with tf.name_scope("train_step"):
        learning_rate = tf.train.exponential_decay(
            LEARNING_RATE_BASE,
            global_step,
            mnist.train.num_examples/BATCH_SIZE, LEARNING_RATE_DECAY,
            staircase=True)

        train_step = tf.train.GradientDescentOptimizer(learning_rate).minimize(loss, global_step=global_step)
        # 在反向传播的过程中更新变量的滑动平均值
        with tf.control_dependencies([train_step, variables_averages_op]):
            train_op = tf.no_op(name='train')
    # 将结果记录进log文件夹中
    writer = tf.summary.FileWriter("log", tf.get_default_graph())

    # 训练模型。
    with tf.Session() as sess:
        tf.global_variables_initializer().run()
        for i in range(TRAINING_STEPS):
            xs, ys = mnist.train.next_batch(BATCH_SIZE)

            if i%1000 == 0:
                # 配置运行时需要记录的信息。
                run_options = tf.RunOptions(trace_level=tf.RunOptions.FULL_TRACE)
                # 运行时记录运行信息的proto。
                run_metadata = tf.RunMetadata()
                # 将配置信息和记录运行信息的proto传入运行的过程，从而记录运行时每一个节点的时间空间开销信息
                _, loss_value, step = sess.run(
                    [train_op, loss, global_step], feed_dict={x: xs, y_: ys},
                    options=run_options, run_metadata=run_metadata)
                writer.add_run_metadata(run_metadata=run_metadata, tag=("tag%d"%i), global_step=i)
                print("After %d training step(s), loss on training batch is %g."%(step, loss_value))
            else:
                _, loss_value, step = sess.run([train_op, loss, global_step], feed_dict={x: xs, y_: ys})

    writer.close()


# #### 3. 主函数。
def main(argv=None):
    mnist = input_data.read_data_sets("../../datasets/MNIST_data", one_hot=True)
    train(mnist)


if __name__ == '__main__':
    main()