Registrations Closing Soon for DevOps Certification Training by CloudxLab | Registrations Closing in

  Enroll Now

Normalising the Features

We shall now normalize the features(pixel values) of train, validation, and test sets with the mean and standard deviation values of the train data.

So first we shall calculate the mean and standard deviation values of the train data.

Then we use these values to normalize the pixel values.

Note:

  • np.newaxis : Used to increase the array dimensionality by 1. Often used when we want to convert a numpy array into one higher dimensionality, like a 1D array to 2d array, or 2D array to 3D array and so on. More here.

  • numpy_array.mean() : Compute the arithmetic mean along the specified axis. More here.

    For example,

    >>import numpy as np
    >>a = np.array(
    [
        [
            [10,11],
            [10,11]
        ],
        [
            [20,21],
            [20,21]
        ],
        [
            [30,31],
            [30,31]
        ]
    ]
    )
    
    >>a.shape
    (3,2,2)
    
    >>print(a.mean())
    20.5
    
    >>print(a.mean(axis=0))
    [[20. 21.]
     [20. 21.]]
    
    >>print(a.mean(axis=1))
    [[10. 11.]
     [20. 21.]
     [30. 31.]]
    
    >>print(a.mean(axis=2))
    [[10.5 10.5]
     [20.5 20.5]
     [30.5 30.5]]
    

    Similar is the case with standard deviation .std(). More here.

INSTRUCTIONS
  • Get the mean X_mean of X_train using mean() along the axis=0.

    X_mean = << your code comes here >>(axis=0, keepdims=True)
    
  • Get the standard deviation X_std of X_train using std() along the axis=0.

    X_std = << your code comes here >>(axis=0, keepdims=True)
    
  • Now, normalize the X_train values as follows:

    X_train = (X_train - X_mean) / X_std
    
  • Similarly, normalize the X_valid values as follows:

    X_valid = << your code comes here >>
    
  • Similarly, normalize the X_test values as follows:

    X_test = << your code comes here >>
    
  • A convolutional layer, which is the first layer of our classifier we are going to build later, expects the input to be 4 dimensional: (batch size, filters or channels, width, height). So let us increase the dimensionality - using np.newaxis - of our data so that everything will be one batch each.

    X_train = X_train[..., np.newaxis]
    X_valid = X_valid[..., np.newaxis]
    X_test = X_test[..., np.newaxis]
    
  • Now let's print the shape of these features in each of the sets:

    print ("Shape of features", X_train.shape, X_valid.shape, X_test.shape)
    

No hints are availble for this assesment

Answer is not availble for this assesment


Note - Having trouble with the assessment engine? Follow the steps listed here

Loading comments...