为Gradient descentoptimizer设置自适应学习率

初始化GradientDescentOptimizer的方式如下：

init = tf.initialize_all_variables()
sess = tf.Session()
sess.run(init)

mse        = tf.reduce_mean(tf.square(out - out_))
train_step = tf.train.GradientDescentOptimizer(0.3).minimize(mse)

方法1：
通过其他普通渐变下降控制学习速率，可以利用以下：tf.train.GradientDescentOptimizer构造函数的learning_rate参数可以是Tensor对象。这允许您在每个步骤中为学习速率计算不同的值，例如：

learning_rate = tf.placeholder(tf.float32, shape=[])
# ...
train_step = tf.train.GradientDescentOptimizer(
    learning_rate=learning_rate).minimize(mse)

sess = tf.Session()

# Feed different values for learning rate to each training step.
sess.run(train_step, feed_dict={learning_rate: 0.1})
sess.run(train_step, feed_dict={learning_rate: 0.1})
sess.run(train_step, feed_dict={learning_rate: 0.01})
sess.run(train_step, feed_dict={learning_rate: 0.01})

方法2：

batch = tf.Variable(0)

learning_rate = tf.train.exponential_decay(
  0.01,                # Base learning rate.
  batch * BATCH_SIZE,  # Current index into the dataset.
  train_size,          # Decay step.
  0.95,                # Decay rate.
  staircase=True)
# Use simple momentum for the optimization.
optimizer = tf.train.MomentumOptimizer(learning_rate,
                                     0.9).minimize(loss,
                                                   global_step=batch)

注意最小化global_step=batch参数。这会告诉优化器在每次训练时都会帮助您增加’batch’参数。

转载自https://stackoverflow.com/questions/33919948/how-to-set-adaptive-learning-rate-for-gradientdescentoptimizer