池化层和全连接层

池化层

max pooling和average pooling

池化区域一般是2*2

全连接层

一般放在最后

#如何将卷积层和池化层结合在一起？
stride

tensorflow中池化层，全连接层函数

1.keras系列
tf.keras.layers.average_pooling2d（
    inputs,
    pool_size,
    strides,
    padding='valid',
    data_format='channels_last',
    name=None
)
#没有变量，没有参数

tf.keras.layers.max_pooling2d（
    inputs,
    pool_size,
    strides,
    padding='valid',
    data_format='channels_last',
    name=None
)

tf.keras.layers.dense(
    inputs,
    units,#输出的维度
    activation=None,
    use_bias=True,
    kernel_initializer=None,
    bias_initializer=tf.zeros_initializer(),
    kernel_regularizer=None,
    bias_regularizer=None,
    activity_regularizer=None,
    kernel_constraint=None,
    bias_constraint=None,
    trainable=True,
    name=None,
    reuse=None
)

2.tf.nn
需要自定义权重
tf.nn.avg_pool(
    value,
    ksize,
    strides,
    padding,
    data_format='NHWC',
    name=None
)

tf.nn.max_pool(
    value,
    ksize,
    strides,
    padding,
    data_format='NHWC',
    name=None
)

#在tf.nn中你能找到全连接层函数吗？
#https://www.tensorflow.org/api_docs/python/tf

3.contrib系列
tf.contrib.layers.avg_pool2d(
    inputs,
    kernel_size,
    stride=2,
    padding='VALID',#默认与卷积层相反
    data_format=DATA_FORMAT_NHWC,
    outputs_collections=None,
    scope=None
)

tf.contrib.layers.max_pool2d(
    inputs,
    kernel_size,
    stride=2,
    padding='VALID',
    data_format=DATA_FORMAT_NHWC,
    outputs_collections=None,
    scope=None
)

tf.contrib.layers.fully_connected(
    inputs,
    num_outputs,
    activation_fn=tf.nn.relu,#
    normalizer_fn=None,
    normalizer_params=None,
    weights_initializer=initializers.xavier_initializer(),
    weights_regularizer=None,
    biases_initializer=tf.zeros_initializer(),
    biases_regularizer=None,
    reuse=None,
    variables_collections=None,#
    outputs_collections=None,#
    trainable=True,
    scope=None
)

#请使用三种不同接口构建如下神经网络
minst数据集
1.卷积层，5*5，输出通道：64
2.最大池化层，2*2
3.卷积层，3*3，输出通道：128
4.平均池化层，2*2
5.全连接层输出通道512
6.全连接层输出通道10
7.softmax层