reduce_xxx

TensorFlow中有一类运算函数具有相同的参数，比如reduce_mean, reduce_min, reduce_max, reduce_sum, reduce_std等。可以对输入tensor按指定轴运算。
以reduce_mean为例，可以写作tf.reduce_mean或者tf.math.reduce_mean:

tf.math.reduce_mean(
input_tensor,
axis = None,
keepdims = False,
name = None
)

• input_tensor: 输入
• axis: 执行运算的轴，如不指定则为全部数据
• keepdims: 运算结果是否保持维度，默认否，降为1维
• name: 可选名称

x = tf.constant([[1., 1.], [2., 2.]])
tf.reduce_mean(x) 
tf.reduce_mean(x, 0) # [1.5, 1.5] 纵轴计算
tf.reduce_mean(x, 1) # [1., 2.] 横轴计算