tf.pad()函数的使用讲解

这个函数主要是对数据矩阵边缘进行填充

import tensorflow as tf
import numpy as np

arr = np.array(np.arange(1, 10)).reshape([3, 3])
print(arr)
b = tf.pad(arr, paddings=[[0, 1], [1, 2]])
sess = tf.InteractiveSession()
print(b.eval())


"""
[[1 2 3]
 [4 5 6]
 [7 8 9]]

--------------------------------
[[0 1 2 3 0 0]
 [0 4 5 6 0 0]
 [0 7 8 9 0 0]
 [0 0 0 0 0 0]]


"""