Tensorflow（五） —— Tensor的broadcast_to操作

1 broadcast_to介绍

1.1 key idea

• insert into 1 dim ahead if needed
• expand dims with size 1 to same size
• 即先升维再扩张到目标数据位数
• broadcast 没有进行数据复制，但却完成相应数据的运算。

1.2how to understand

• when it has no axis, create a new concept(axis)
• when it has dim of size 1,

1.3 why broadcasting

• for real demading
• memory consumption

1.4 broadcastable

• Match from last dim 最右边必须对齐 位数相同或者为1
• if current dim=1,expand to same
• if either has no dim, insert one dim and expand to same
• otherwise, Not broadcastable

[4,32,14,14]
[2,32,14,14]
not broadcastable

2 隐式

"""
a = tf.random.uniform([4,28,28,3])

b = tf.constant(5.)
print("a+b:",(a+<