C4- confused about multi dimensional array representation and coding

Course Q&A Deep Learning Specialization Convolutional Neural Networks
1

Hi, I am confused about multi-dimensional array representation in reality vs python code.
for example:
This is called a 6x6x3 image (array)
image

however, when executing a code like

x= np.ones((6, 6, 3))
print(x)

I get this output

[[[1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]]

 [[1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]]

 [[1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]]

 [[1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]]

 [[1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]]

 [[1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]]]

This is equivalent to a 6x3x6 image.

And if we add another dimension to represent a dataset of m RGB images of the same size I believe it would be represented as 6x6x3xm but to be defined in python the shape should be (m, 3, 6, 6).
So, In the first assignment

A_prev = np.random.randn(2, 5, 5, 3)

should be representing a dataset with two 5-channeled 5x3 images, right?
if yes, then this line

    # Retrieve dimensions from the input shape
    (m, n_H_prev, n_W_prev, n_C_prev) = A_prev.shape

shouldn’t be

    # Retrieve dimensions from the input shape
    (m,  n_C_prev, n_H_prev, n_W_prev) = A_prev.shape

?
Sorry for the lengthy question.

2

How you view the array does not have any effect on the actual order of the dimensions, right? The order of the dimensions is set by intention. There are choices, but Prof Ng has made the choices for us. And the choice is that the order of the dimensions is as shown in that line of code:

1 Like