tensorflow中的正则化函数在_『TensorFlow』正则化添加方法整理

一、基础正则化函数

tf.contrib.layers.l1_regularizer(scale, scope=None)

返回一个用来执行L1正则化的函数,函数的签名是func(weights).

参数:

scale: 正则项的系数.

scope: 可选的scope name

tf.contrib.layers.l2_regularizer(scale, scope=None)

先看看tf.contrib.layers.l2_regularizer(weight_decay)都执行了什么：

import tensorflow as tf

sess=tf.Session()

weight_decay=0.1

tmp=tf.constant([0,1,2,3],dtype=tf.float32)

"""

l2_reg=tf.contrib.layers.l2_regularizer(weight_decay)

a=tf.get_variable("I_am_a",regularizer=l2_reg,initializer=tmp)

"""

#**上面代码的等价代码

a=tf.get_variable("I_am_a",initializer=tmp)

a2=tf.reduce_sum(a*a)*weight_decay/2;

a3=tf.get_variable(a.name.split(":")[0]+"/Regularizer/l2_regularizer",initializer=a2)