定义权重正则化损失和如何规定要计算梯度的变量
权重正则化损失
在使⽤tf.get_variable()和tf.variable_scope()的时候，你会发现，它们俩中有regularizer形参.如果传⼊这个参数的话，那么variable_scope内的weights的正则化损失，或者weights 的正则化损失就会被添加到GraphKeys.REGULARIZATION_LOSSES中.
⽰例
import tensorflow as tf
ib import layers
regularizer = layers.l1_regularizer(0.1)
with tf.variable_scope('var', initializer=tf.random_normal_initializer(), regularizer=regularizer):
weight = tf.get_variable('weight', shape=[8], s_initializer())
with tf.variable_scope('var2', initializer=tf.random_normal_initializer(), regularizer=regularizer):
weight2 = tf.get_variable('weight', shape=[8], s_initializer())
regularization_loss = tf.reduce__collection(tf.GraphKeys.REGULARIZATION_LOSSES))
正则化残差
optimize_loss = tf.train.AdamOptimizer().minimize(loss + sum(regularization_loss), var_list=ae_vars)
如何规定要计算梯度的变量
t_vars = tf.trainable_variables()  //获取全部变量
ae_vars = [var for var in t_vars if 'autoencoder' in var.name]  // 取出要计算机梯度的变量
optimize_loss = tf.train.AdamOptimizer().minimize(loss + sum(regularization_loss), var_list=ae_vars)

版权声明：本站内容均来自互联网，仅供演示用，请勿用于商业和其他非法用途。如果侵犯了您的权益请与我们联系QQ:729038198，我们将在24小时内删除。