tf.get_collection()

此函数有两个参数，key和scope。

Args:

• 1.key: The key for the collection. For example, the GraphKeys class contains many standard names for collections.
• 2.scope: (Optional.) If supplied, the resulting list is filtered to include only items whose name attribute matches using re.match. Items without a name attribute are never returned if a scope is supplied and the choice or re.match means that a scope without special tokens filters by prefix.

举个例子：

# 在'My-TensorFlow-tutorials-master/02 CIFAR10/cifar10.py'代码中

  variables = tf.get_collection(tf.GraphKeys.VARIABLES)
  for i in variables:
  print(i)

>>>   
      
      
      
      
      
      
      
      
      

tf.get_collection会列出key里所有的值。

---

进一步地：

tf.GraphKeys的点后可以跟很多类，
比如VARIABLES类（包含所有variables），
比如REGULARIZATION_LOSSES。

具体tf.get_collection(tf.GraphKeys.REGULARIZATION_LOSSES)的使用：

def easier_network(x, reg):
  """ A network based on tf.contrib.learn, with input `x`. """
  with tf.variable_scope('EasyNet'):
     out = layers.flatten(x)
     out = layers.fully_connected(out,
                                  num_outputs=200,
                                  weights_initializer = layers.xavier_initializer(uniform=True),
                                  weights_regularizer = layers.l2_regularizer(scale=reg),
                                  activation_fn = tf.nn.tanh)
     out = layers.fully_connected(out,
                                  num_outputs=200,
                                  weights_initializer = layers.xavier_initializer(uniform=True),
                                  weights_regularizer = layers.l2_regularizer(scale=reg),
                                  activation_fn = tf.nn.tanh)
     out = layers.fully_connected(out,
                                  num_outputs=10, # Because there are ten digits!
                                  weights_initializer = layers.xavier_initializer(uniform=True),
                                  weights_regularizer = layers.l2_regularizer(scale=reg),
                                  activation_fn = None)
     return out

 def main(_):
  mnist = input_data.read_data_sets(FLAGS.data_dir, one_hot=True)
  x = tf.placeholder(tf.float32, [None, 784])
  y_ = tf.placeholder(tf.float32, [None, 10])

  # Make a network with regularization
  y_conv = easier_network(x, FLAGS.regu)
  weights = tf.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES, 'EasyNet')
  print("")
  for w in weights:
     shp = w.get_shape().as_list()
     print("- {} shape:{} size:{}".format(w.name, shp, np.prod(shp)))
     print("")
     reg_ws = tf.get_collection(tf.GraphKeys.REGULARIZATION_LOSSES, 'EasyNet')
  for w in reg_ws:
     shp = w.get_shape().as_list()
     print("- {} shape:{} size:{}".format(w.name, shp, np.prod(shp)))
     print("")

  # Make the loss function `loss_fn` with regularization.
  cross_entropy = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels=y_, logits=y_conv))
  loss_fn = cross_entropy + tf.reduce_sum(reg_ws)
  train_step = tf.train.AdamOptimizer(1e-4).minimize(loss_fn)

main()

>>>   - EasyNet/fully_connected/weights:0 shape:[784, 200] size:156800
      - EasyNet/fully_connected/biases:0 shape:[200] size:200
      - EasyNet/fully_connected_1/weights:0 shape:[200, 200] size:40000
      - EasyNet/fully_connected_1/biases:0 shape:[200] size:200
      - EasyNet/fully_connected_2/weights:0 shape:[200, 10] size:2000
      - EasyNet/fully_connected_2/biases:0 shape:[10] size:10

      - EasyNet/fully_connected/kernel/Regularizer/l2_regularizer:0 shape:[] size:1.0
      - EasyNet/fully_connected_1/kernel/Regularizer/l2_regularizer:0 shape:[] size:1.0
      - EasyNet/fully_connected_2/kernel/Regularizer/l2_regularizer:0 shape:[] size:1.0

据:

for w in reg_ws:
     shp = ....

这段代码的输出可知，
在图上的所有regularization都会集中保存到tf.GraphKeys.REGULARIZATION_LOSSES去。

关于collection的详情请参见：
http://blog.csdn.net/shenxiaolu1984/article/details/52815641

关于tf.GraphKeys.REGULARIZATION_LOSSES的详情参见：
https://gxnotes.com/article/178205.html