visualize_data.py

'''
Render the first 9 images of the MNIST data in a grid.
Just to show that these are indeed low-resolution
handwriting samples.
'''
import matplotlib.pyplot as plt
import mnist_loader

def load_mnist_pickle_file():
 '''
 Load the MNIST dataset from a local copy of the file.
 Can also get bigger/smaller datasets from here:
 http://yann.lecun.com/exdb/mnist/

 Can train (fit) the model on the smaller set and predict on the larger
 '''
 print ("loading MNIST training data...")
 training_data, _, test_data = mnist_loader.load_data(pickle_filename='./mnist.pkl.gz')
 return (training_data[0], training_data[1]), (test_data[0], test_data[1])


# x_train will be the training images
# y_train will be the labels for each item
def visualize():
 _ , (x_test, y_test) = load_mnist_pickle_file()

 for i in range(9):
 plt.subplot(330 + 1 + i)
 plt.imshow(x_test[i].reshape((28,28)), cmap=plt.get_cmap('gray'))
 plt.show()

def visualizer_error(error_indices: list[int], test_data):
 # Get the first 9 indices where predictions != y_test
 # Visualize first nine incorrect images
 for count, idx in enumerate(error_indices):
 plt.subplot(330 + 1 + count)
 plt.imshow(test_data[idx].reshape((28,28)), cmap=plt.get_cmap('gray'))
 plt.show()