TensorFlow 中的批量归一化初始值设定项 [英] Batch normalization initializer in TensorFlow

问题描述

在 TensorFlow 中，批量归一化参数包括 beta、gamma、moving mean 和 moving variance.但是，为了初始化这些参数，tf.contrib.layers.batch_norm(*args, **kwargs) 中只有一个参数叫做 param_initializers，根据它包含的文档beta、gamma、moving mean 和 moving variance 的可选初始化器.

In TensorFlow, batch normalization parameters include beta, gamma, moving mean, and moving variance. However, for initializing these parameters there is only one argument in tf.contrib.layers.batch_norm(*args, **kwargs) called param_initializers which according to the documents it contains optional initializers for beta, gamma, moving mean and moving variance.

我们如何使用param_initializers来初始化这些参数?

How can we use param_initializers to initialize these parameters?

推荐答案

这里是如何使用 使用 Tensorflow 1.0 进行批量标准化:

import tensorflow as tf
batch_normalization = tf.layers.batch_normalization

... (define the network)
net = batch_normalization(net)
... (define the network)

如果你想设置参数，就这样做:

If you want to set parameters, just do it like this:

net = batch_normalization(net, 
                          beta_initializer=tf.zeros_initializer(), 
                          moving_variance_initializer=tf.ones_initializer())

*args, **kwargs

这是传递任意多个非关键字参数args和任意多个关键字参数kwargs的python方式.例如:

*args, **kwargs

This is the python way to pass arbitrary many non-keyword arguments args and arbitrary many keyword arguments kwargs. For example:

def test(*args, **kwargs):
    print("#" * 80)
    print(args)
    print("#" * 80)
    print(kwargs)

test(1, 2, 42, 3.141, 'foo', a=7, b=3, c='bla')

给予

################################################################################
(1, 2, 42, 3.141, 'foo')
################################################################################
{'a': 7, 'c': 'bla', 'b': 3}

这篇关于TensorFlow 中的批量归一化初始值设定项的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！