神经网络优化算法如何选择Adam，SGD
之前在tensorflow上和caffe上都折腾过CNN用来做视频处理，在学习tensorflow例子的时候代码里面给的优化方案默认很多情况下都是直接用的AdamOptimizer优化算法，如下：
optimizer = tf.train.AdamOptimizer(learning_rate=lr).minimize(cost)但是在使用caffe时solver里面一般都用的SGD+momentum，如下：
base_lr: 0.0001
momentum: 0.9
weight_decay: 0.0005
lr_policy: "step"加上最近看了一篇文章：The Marginal Value of Adaptive Gradient Methods
in Machine Learning文章链接，文中也探讨了在自适应优化算法：AdaGrad, RMSProp, and Adam和SGD算法性能之间的比较和选择，因此在此搬一下结论和感想。
Abstract
经过本文的实验，得出最重要的结论是：
We observe that the solutions found by adaptive methods generalize worse (often significantly worse) than SGD, even when these solutions have better training performance. These
results suggest that practitioners should reconsider the use of adaptive methods to train neural
networks翻译一下就是自适应优化算法通常都会得到比SGD算法性能更差（经常是差很多）的结果，尽管自适应优化算法在训练时会表现的比较好，因此使用者在使用自适应优化算法时需要慎重考虑！（终于知道为啥CVPR的paper全都用的SGD了，而不是用理论上最diao的Adam）
Introduction
作者继续给了干货结论：
Our experiments reveal three primary findings.
First,
with the same amount of hyperparameter tuning, SGD and SGD with momentum outperform
adaptive methods on the development/test set across all evaluated models and tasks. This is
true even when the adaptive methods achieve the same training loss or lower than non-adaptive
methods. Second, adaptive methods often display faster initial progress on the training set, but
their performance quickly plateaus on the development/test set. Third, the same amount of tuning
was required for all methods, including adaptive methods. This challenges the conventional wisdom
that adaptive methods require less tuning. 翻译：
1：用相同数量的超参数来调参，SGD和SGD +momentum 方法性能在测试集上的额误差好于所有的自适应优化算法，尽管有时自适应优化算法在训练集上的loss更小，但是他们在测试集上的loss却依然比SGD方法高，
2：自适应优化算法 在训练前期阶段在训练集上收敛的更快，但是在测试集上这种有点遇到了瓶颈。
3：所有方法需要的迭代次数相同，这就和约定俗成的默认自适应优化算法 需要更少的迭代次数的结论相悖！
Conclusion
贴几张作者做的实验结果图：
可以看到SGD在训练前期loss下降并不是最快的，但是在test set上的Perplexity 困惑度（这里写链接内容）是最小的。
在tensorflow中使用SGD算法：（参考）
# global_step
training_iters=len(data_config['train_label'])
global_step=training_iters*model_config['n_epoch']
decay_steps=training_iters*1
#global_step = tf.Variable(0, name = 'global_step', trainable=False)
lr=tf.train.exponential_decay(learning_rate=model_config['learning_rate'],
global_step=global_step, decay_steps=decay_steps, decay_rate=0.1, staircase=False, name=None)
optimizer=tf.train.GradientDescentOptimizer(lr).minimize(cost,var_list=network.all_params)