在第16节，我们建立了最基本的深度学习网络，并且进行了训练。在第19节，参照官方例子，给出了实时评估训练结果的例子。训练到 30000 步时（batch=100），正确率达到了 70%。本节，将分析一下官方例子的代码。
官方代码如何建立网络
官方给出的卷积神经网络部分的结构非常清楚。
卷积1 -> 池化1 -> lrn层1 -> 卷积2 -> lrn层2 -> 池化2 -> 全连接层 x3
具体代码如下
def inference(images):
"""Build the CIFAR-10 model.
Args:
images: Images returned from distorted_inputs() or inputs().
Returns:
Logits.
"""
# We instantiate all variables using tf.get_variable() instead of
# tf.Variable() in order to share variables across multiple GPU training runs.
# If we only ran this model on a single GPU, we could simplify this function
# by replacing all instances of tf.get_variable() with tf.Variable().
#
# conv1
with tf.variable_scope('conv1') as scope:
kernel = _variable_with_weight_decay('weights',
shape=[5, 5, 3, 64],
stddev=5e-2,
wd=0.0)
conv = tf.nn.conv2d(images, kernel, [1, 1, 1, 1], padding='SAME')
biases = _variable_on_cpu('biases', [64], tf.constant_initializer(0.0))
pre_activation = tf.nn.bias_add(conv, biases)
conv1 = tf.nn.relu(pre_activation, name=scope.name)
_activation_summary(conv1)
# pool1
pool1 = tf.nn.max_pool(conv1, ksize=[1, 3, 3, 1], strides=[1, 2, 2, 1],
padding='SAME', name='pool1')
# norm1
norm1 = tf.nn.lrn(pool1, 4, bias=1.0, alpha=0.001 / 9.0, beta=0.75,
name='norm1')
# conv2
with tf.variable_scope('conv2') as scope:
kernel = _variable_with_weight_decay('weights',
shape=[5, 5, 64, 64],
stddev=5e-2,
wd=0.0)
conv = tf.nn.conv2d(norm1, kernel, [1, 1, 1, 1], padding='SAME')
biases = _variable_on_cpu('biases', [64], tf.constant_initializer(0.1))
pre_activation = tf.nn.bias_add(conv, biases)
conv2 = tf.nn.relu(pre_activation, name=scope.name)
_activation_summary(conv2)
# norm2
norm2 = tf.nn.lrn(conv2, 4, bias=1.0, alpha=0.001 / 9.0, beta=0.75,
name='norm2')
# pool2
pool2 = tf.nn.max_pool(norm2, ksize=[1, 3, 3, 1],
strides=[1, 2, 2, 1], padding='SAME', name='pool2')
# local3
with tf.variable_scope('local3') as scope:
# Move everything into depth so we can perform a single matrix multiply.
reshape = tf.reshape(pool2, [FLAGS.batch_size, -1])
dim = reshape.get_shape()[1].value
weights = _variable_with_weight_decay('weights', shape=[dim, 384],
stddev=0.04, wd=0.004)
biases = _variable_on_cpu('biases', [384], tf.constant_initializer(0.1))
local3 = tf.nn.relu(tf.matmul(reshape, weights) + biases, name=scope.name)
_activation_summary(local3)
# local4
with tf.variable_scope('local4') as scope:
weights = _variable_with_weight_decay('weights', shape=[384, 192],
stddev=0.04, wd=0.004)
biases = _variable_on_cpu('biases', [192], tf.constant_initializer(0.1))
local4 = tf.nn.relu(tf.matmul(local3, weights) + biases, name=scope.name)
_activation_summary(local4)
# linear layer(WX + b),
# We don't apply softmax here because
# tf.nn.sparse_softmax_cross_entropy_with_logits accepts the unscaled logits
# and performs the softmax internally for efficiency.
with tf.variable_scope('softmax_linear') as scope:
weights = _variable_with_weight_decay('weights', [192, NUM_CLASSES],
stddev=1/192.0, wd=0.0)
biases = _variable_on_cpu('biases', [NUM_CLASSES],
tf.constant_initializer(0.0))
softmax_linear = tf.add(tf.matmul(local4, weights), biases, name=scope.name)
_activation_summary(softmax_linear)
return softmax_linear
这里使用了如下技巧：
- 对 weights 进行了 L2 的正则化
- 在每个卷积-最大池化层后面使用了 lrn 层，增加了泛化能力
初始化 weight 参数时，用到如下方法：
def _variable_with_weight_decay(name, shape, stddev, wd):
var = _variable_on_cpu(
name,
shape,
tf.truncated_normal_initializer(stddev=stddev, dtype=dtype))
if wd is not None:
weight_decay = tf.multiply(tf.nn.l2_loss(var), wd, name='weight_loss')
tf.add_to_collection('losses', weight_decay)
return var
这里初始化 weight 时，利用了 tf.truncated_normal 截断的正态分布初始化，但是还会给 weight 加了一个 L2 的loss，相当于多做了一个 L2 的正则化处理，目的是为了减少过拟合。以上方法，使用了 w1 控制 L2 loss 的大小，用 tf.nn.l2_loss 方法计算了 weight 的 L2 loss，然后乘以w1 得到最终的输出。
一般来说，L1 正则会制造稀疏的特征，大部分无用特征的权重会被置零，而 L2 正则则会让特征的权重比较平均。
注意到inference中的池化层形状为 3x3，步长为 2x2，让步长小于池化层的尺寸，可以增加数据的丰富性。使用 tf.nn.lrn 方法处理数据则主要是模仿了生物神经系统的侧边抑制机制，目的是抑制反馈较小的神经元，放大反馈较大的神经元，从而增强模型的泛化能力。lrn 层适合 relu 这种没有上限边界的激活函数，因为它会从附近的多个卷积核的响应中挑选较大的反馈，但是不适合 sigmoid 这种有固定边界并且能抑制过大值的激活函数。
在最后的全连接层后，网络就结束了，没有使用 softmax。（softmax 计算官方例子放在 loss 里计算了，而计算softmax 的目的就是为了计算 loss，因此放在 loss 里计算时合适的。）
官方例子如何建立 loss
建立 loss 非常简单，主要使用了 tensorflow 的 sparse_softmax_cross_entropy_with_logits 方法，这个方法我们在第17节有所介绍。代码如下
def loss(logits, labels):
"""Add L2Loss to all the trainable variables.
Add summary for "Loss" and "Loss/avg".
Args:
logits: Logits from inference().
labels: Labels from distorted_inputs or inputs(). 1-D tensor
of shape [batch_size]
Returns:
Loss tensor of type float.
"""
# Calculate the average cross entropy loss across the batch.
labels = tf.cast(labels, tf.int64)
cross_entropy = tf.nn.sparse_softmax_cross_entropy_with_logits(
labels=labels, logits=logits, name='cross_entropy_per_example')
cross_entropy_mean = tf.reduce_mean(cross_entropy, name='cross_entropy')
tf.add_to_collection('losses', cross_entropy_mean)
# The total loss is defined as the cross entropy loss plus all of the weight
# decay terms (L2 loss).
return tf.add_n(tf.get_collection('losses'), name='total_loss')
官方例子如何建立优化器进行训练（train）
官方提供的卷积神经网络使用的优化器是最基本的梯度下降优化器，只不过在学习率上有所改动。
def train(total_loss, global_step):
"""Train CIFAR-10 model.
Create an optimizer and apply to all trainable variables. Add moving
average for all trainable variables.
Args:
total_loss: Total loss from loss().
global_step: Integer Variable counting the number of training steps
processed.
Returns:
train_op: op for training.
"""
# Variables that affect learning rate.
num_batches_per_epoch = NUM_EXAMPLES_PER_EPOCH_FOR_TRAIN / FLAGS.batch_size
decay_steps = int(num_batches_per_epoch * NUM_EPOCHS_PER_DECAY)
# Decay the learning rate exponentially based on the number of steps.
lr = tf.train.exponential_decay(INITIAL_LEARNING_RATE,
global_step,
decay_steps,
LEARNING_RATE_DECAY_FACTOR,
staircase=True)
tf.summary.scalar('learning_rate', lr)
# Generate moving averages of all losses and associated summaries.
loss_averages_op = _add_loss_summaries(total_loss)
# Compute gradients.
with tf.control_dependencies([loss_averages_op]):
opt = tf.train.GradientDescentOptimizer(lr)
grads = opt.compute_gradients(total_loss)
# Apply gradients.
apply_gradient_op = opt.apply_gradients(grads, global_step=global_step)
# Add histograms for trainable variables.
for var in tf.trainable_variables():
tf.summary.histogram(var.op.name, var)
# Add histograms for gradients.
for grad, var in grads:
if grad is not None:
tf.summary.histogram(var.op.name + '/gradients', grad)
# Track the moving averages of all trainable variables.
variable_averages = tf.train.ExponentialMovingAverage(
MOVING_AVERAGE_DECAY, global_step)
variables_averages_op = variable_averages.apply(tf.trainable_variables())
with tf.control_dependencies([apply_gradient_op, variables_averages_op]):
train_op = tf.no_op(name='train')
return train_op
可以看出学习率并不是一个常数，它会随着训练次数的增加变动。
# Decay the learning rate exponentially based on the number of steps.
lr = tf.train.exponential_decay(INITIAL_LEARNING_RATE,
global_step,
decay_steps,
LEARNING_RATE_DECAY_FACTOR,
staircase=True)
据官方数据，这样的训练可以提升正确率，最终可以达到接近 86% 的正确率。当然，正确率的提高也得益于 L2 正则即 lrn 层的使用。另外，数据增强（随机裁剪图片，亮度调节等）官方例子也是使用了的，在第16节建立网络时，就参考了这一点。
