将数组和张量转换为矩阵，然后执行矩阵运算

Question

I have two arrays of size (128,) and a third one of size (784, 128):

array1.shape()
out: (128,)
array2.shape()
out: (128,)
array3.shape()
out: (784,128)

They have the same data type but the dtype() output is different:

array1.dtype
out: float32
array2.dtype
out: <dtype: 'float32'>
array2.dtype
out: <dtype: 'float32'>

And they belong to different classes:

type(array1)
out: <class 'numpy.ndarray'>
type(array2)
out: <class 'tensorflow.python.ops.resource_variable_ops.ResourceVariable'>
type(array3)
out: <class 'tensorflow.python.ops.resource_variable_ops.ResourceVariable'>

I want to perform the following matrix operation:

(array1 - array2) * array3.T

Where T is the transpose of array3.

Lastly, the output matrix (which is [784 * 1]) needs to be reshaped to become a uint8 array of shape 28 * 28 so I can plot that output on a matplotlip .

Can anyone help me to convert the arrays into matrices first. Then Transpose the third array properly. Finally, reshape the output to become an uint8 array of size 28 * 28.

I am working with tensorflow and keras in python.

Answer 1

Cast array1 as np.asmatrix()

array1 = np.asmatrix(array1)

For array2, change it to numpy array fist then cast it as a matrix:

array2 = array2.numpy()
array2 = np.asmatrix(array2)

For array3, change it to numpy array first then cast it as a matrix. Lastly, transpose that matrix:

array3 = array3.numpy()
array3 = np.asmatrix(array3)
array3 = np.transpose(array3)

Finally, apply the matrix operation:

result = (array1 - array2) * array3

To plot the output, use the following:

result = np.array(result)
plt.imshow(result.reshape(28, 28), cmap='gray')
plt.show()