tf.cumsum

简介

• 累计求和

函数

cumsum(
    x,
    axis=0,
    exclusive=False,
    reverse=False,
    name=None
)

exclusive

• True：排除第一项
• False：不排除第一项

reverse

• True：逆向求和
• False：顺序求和

示例

a = [[1 ,2, 3], 
     [4, 5, 6], 
     [7, 8, 9]]
# axis=0
sum1 = tf.cumsum(a, axis=0)

sum1 = [[ 1,  2,  3],
        [ 5,  7,  9],
        [12, 15, 18]]

# axis=1
sum2 = tf.cumsum(a, axis=1)

sum2 = [[ 1,  3,  6],
        [ 4,  9, 15],
        [ 7, 15, 24]]

# exclusive=False
sum3 = tf.cumsum(a, exclusive=False)

sum3 = [[ 1,  2,  3],
        [ 5,  7,  9],
        [12, 15, 18]]

# exclusive=True
sum4 = tf.cumsum(a, exclusive=True)

sum4= [[0, 0, 0],
       [1, 2, 3],
       [5, 7, 9]]


# reverse=True
sum5 = tf.cumsum(a, reverse=True)

sum5 = [[12, 15, 18],
        [11, 13, 15],
        [ 7,  8,  9]]

# exclusive=True, reverse=True
sum6 = tf.cumsum(a, exclusive=True, reverse=True)

sum6 = [[11, 13, 15],
        [ 7,  8,  9],
        [ 0,  0,  0]]

参考

• https://blog.csdn.net/YiRanNingJing/article/details/79451786