- html - 出于某种原因,IE8 对我的 Sass 文件中继承的 html5 CSS 不友好?
- JMeter 在响应断言中使用 span 标签的问题
- html - 在 :hover and :active? 上具有不同效果的 CSS 动画
- html - 相对于居中的 html 内容固定的 CSS 重复背景?
我目前正在学习 CS231 作业,我意识到一些令人困惑的事情。在计算梯度时,当我第一次 reshape x 然后得到转置时,我得到了正确的结果。
x_r=x.reshape(x.shape[0],-1)
dw= x_r.T.dot(dout)
dw = x.reshape(-1,x.shape[0]).dot(dout)
最佳答案
虽然您的两种方法都会产生相同形状的数组,但由于 numpy 读取/写入元素的方式,元素的顺序会有所不同。默认情况下,reshape
使用类似 C 的索引顺序,这意味着元素被读取/写入,最后一个轴索引变化最快,回到第一个轴索引变化最慢(取自 documentation )。
这是一个在实践中意味着什么的例子。让我们假设以下数组 x
:
x = np.asarray([[[1, 2], [3, 4], [5, 6]], [[7, 8], [9, 10], [11, 12]]])
print(x.shape) # (2, 3, 2)
print(x)
# output
[[[ 1 2]
[ 3 4]
[ 5 6]]
[[ 7 8]
[ 9 10]
[11 12]]]
现在让我们通过以下两种方式 reshape 这个数组:
opt1 = x.reshape(x.shape[0], -1)
opt2 = x.reshape(-1, x.shape[0])
print(opt1.shape) # outptu: (2, 6)
print(opt2.shape) # output: (6, 2)
print(opt1)
# output:
[[ 1 2 3 4 5 6]
[ 7 8 9 10 11 12]]
print(opt2)
# output:
[[ 1 2]
[ 3 4]
[ 5 6]
[ 7 8]
[ 9 10]
[11 12]]
reshape
首先推断出新数组的形状,然后返回一个 View ,在该 View 中以类似 C 的索引顺序读取元素。
opt1
上举例说明这一点: 由于原始数组
x
有 12 个元素,它推断新数组
opt1
形状必须为
(2, 6)
(因为 2*6=12)。现在,
reshape
返回一个 View ,其中:
opt1[0][0] == x[0][0][0]
opt1[0][1] == x[0][0][1]
opt1[0][2] == x[0][1][0]
opt1[0][3] == x[0][1][1]
opt1[0][4] == x[0][2][0]
opt1[0][5] == x[0][2][1]
opt1[1][0] == x[1][0][0]
...
opt1[1][5] == x[1][2][1]
所以如上所述,最后一个轴索引变化最快,第一个轴索引变化最慢。同样,
opt2
的输出将被计算。
opt1 = opt1.T
print(opt1.shape) # output: (6, 2)
print(opt1)
# output:
[[ 1 7]
[ 2 8]
[ 3 9]
[ 4 10]
[ 5 11]
[ 6 12]]
显然,由于元素排序,这两种方法不会产生相同的数组,即使它们具有相同的形状。
关于python - 只是 reshape 和 reshape 和获得转置之间的区别?,我们在Stack Overflow上找到一个类似的问题: https://stackoverflow.com/questions/68700008/
如何检查字符串是否被 reshape ?示例:“aab”返回 0,因为“a”无法 reshape 为该字符串或任何其他更短的字符串。 另一个例子是“aabbaab”返回 1,因为“aabb”可以被 r
我无法清楚地理解theano的reshape。我有一个形状的图像矩阵: [batch_size, stack1_size, stack2_size, height, width] ,其中有 s
如何检查字符串是否被 reshape ?示例:“aab”返回 0,因为“a”无法 reshape 为该字符串或任何其他更短的字符串。 另一个例子是“aabbaab”返回 1,因为“aabb”可以被 r
这是原始数据 a=[[1,2,3,4,5,6], [7,8,9,10,11,12]] 我想把它转换成这样的格式: b=[[1,2,3,7,8,9], [4,5,6,10,11,12]] a
我目前正在学习 CS231 作业,我意识到一些令人困惑的事情。在计算梯度时,当我第一次 reshape x 然后得到转置时,我得到了正确的结果。 x_r=x.reshape(x.shape[0],-1
这个问题在这里已经有了答案: Reshaping multiple sets of measurement columns (wide format) into single columns (lon
我有一个包含超过 1500 列的宽格式数据集。由于许多变量都是重复的,我想将其 reshape 为长形式。然而,r 抛出一个错误: Error in guess(varying) : Failed
我有一个长格式的数据框狗,我正在尝试使用 reshape() 函数将其重新格式化为宽格式。目前看起来是这样的: dogid month year trainingtype home scho
这个问题在这里已经有了答案: how to reshape an N length vector to a 3x(N/3) matrix in numpy using reshape (1 个回答)
我对 ndarray.reshape 的结构有疑问.我读过 numpy.reshape()和 ndarray.reshape是 python 中用于 reshape 数组的等效命令。 据我所知,num
所以这是我的麻烦:我想将一个长格式的数据文件改成宽格式。但是,我没有唯一的“j”变量;长格式文件中的每条记录都有几个关键变量。 例如,我想这样做: | caseid | gender | age |
Whis 这个数据框, df df id parameter visit value sex 1 01 blood V1 1 f 2 01 saliva V
我有一个列表,其中包含几个不同形状的 numpy 数组。我想将这个数组列表 reshape 为一个 numpy 向量,然后更改向量中的每个元素,然后将其 reshape 回原始数组列表。 例如: 输入
我有一个形状为 (1800,144) 的数组 (a) 其中 a[0:900,:] 都是实数,后半部分数组 a[900:1800,:] 全部为零。我想把数组的后半部分水平地放在前半部分旁边,然后将它们推
我有一个如下所示的数组: array([[0, 0, 0, 0, 0, 0, 0, 0], [1, 1, 1, 1, 1, 1, 1, 1], [2, 2, 2, 2, 2
我正在创建一个 tf.Variable(),然后使用该变量创建一个简单的函数,然后我使用 tf.reshape() 展平原始变量,然后我在函数和展平变量之间使用了 tf.gradients()。为什么
我有一个名为 data 的数据框,我试图从中识别任何异常价格。 数据框头部看起来像: Date Last Price 0 29/12/2017 487.74 1 28/
我有一个 float vec 数组,我想对其进行 reshape vec.shape >>> (3,) len(vec[0]) # all 3 rows of vec have 150 columns
tl;dr 我可以在不使用 numpy.reshape 的情况下将 numpy 数组的 View 从 5x5x5x3x3x3 reshape 为 125x1x1x3x3x3 吗? 我想对一个体积(大小
set.seed(123)data <- data.frame(ID = 1:10, weight_hus = rnorm(10, 0, 1),
我是一名优秀的程序员,十分优秀!