A0_MNIST_CNN_Keras_TypeA.py

'''
Data Engineering
'''

'''
D1. Import Libraries for Data Engineering
'''
import tensorflow as tf

import matplotlib.pyplot as plt

'''
D2. Load MNIST data / Only for Toy Project
'''

# print(tf.__version__)
## MNIST Dataset #########################################################
mnist = tf.keras.datasets.mnist
# class_names = ['0', '1', '2', '3', '4', '5', '6', '7', '8', '9']
##########################################################################

## Fashion MNIST Dataset #################################################
# mnist = tf.keras.datasets.fashion_mnist
# class_names = ['T-shirt/top', 'Trouser', 'Pullover', 'Dress', 'Coat', 'Sandal', 'Shirt', 'Sneaker', 'Bag', 'Ankle boot']
##########################################################################
(X_train, Y_train), (X_test, Y_test) = mnist.load_data()

# Change data type as float. If it is int type, it might cause error 
'''
D3. Data Preprocessing
'''
# Normalizing
X_train, X_test = X_train / 255.0, X_test / 255.0

print(Y_train[0:10])
print(X_train.shape)

'''
D4. EDA(? / Exploratory data analysis)
'''

# plot first few images
for i in range(9):
    # define subplot
    plt.subplot(330 + 1 + i)
    # plot raw pixel data
    plt.imshow(X_train[i], cmap=plt.get_cmap('gray'))
    # if you want to invert color, you can use 'gray_r'. this can be used only for MNIST, Fashion MNIST not cifar10
    # pyplot.imshow(trainX[i], cmap=pyplot.get_cmap('gray_r'))
    
# show the figure
plt.show()

'''
Model Engineering
'''

'''
M1. Import Libraries for Model Engineering
'''
from tensorflow.keras.layers import Dense, Flatten, Conv2D, MaxPool2D, Dropout
from tensorflow.keras import Model, Sequential
'''
M2. Set Hyperparameters
'''

hidden_size = 256
output_dim = 10      # output layer dimensionality = num_classes
EPOCHS = 30

'''
M3. Build NN model
'''
# in the case of Keras or TF2, type shall be [image_size, image_size, 1]
model = Sequential([
    Conv2D(filters=64, kernel_size=3, activation=tf.nn.relu, padding='SAME',input_shape=(28, 28, 1)),
    MaxPool2D(padding='SAME'),
    Conv2D(filters=128, kernel_size=3, activation=tf.nn.relu, padding='SAME'),
    MaxPool2D(padding='SAME'),
    Conv2D(filters=256, kernel_size=3, activation=tf.nn.relu, padding='SAME'),
    MaxPool2D(padding='SAME'),
    Flatten(),
    Dense(hidden_size, activation='relu'),
    Dropout(0.2),
    Dense(output_dim, activation='softmax')
])

model.summary()

'''
M4. Optimizer
'''
# Optimizer can be included at model.compile

'''
M5. Model Compilation - model.compile
'''

model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])

'''
M6. Train and Validation - `model.fit`
'''

model.fit(X_train, Y_train, epochs=EPOCHS, verbose=1)

'''
M7. Assess model performance
'''
loss, acc = model.evaluate(X_test, Y_test, verbose=0)
print('test_loss: {:.3f}, test_acc: {:.3f}'.format(
    loss,
    acc
))

'''
M7. [Opt] Assess model performance
'''
model.evaluate(X_test,  Y_test, verbose=2)              


'''

Keras Display Options

verbose default is 1

verbose=0 (silent)

verbose=1 (progress bar)

Train on 186219 samples, validate on 20691 samples
Epoch 1/2
186219/186219 [==============================] - 85s 455us/step - loss: 0.5815 - acc: 
0.7728 - val_loss: 0.4917 - val_acc: 0.8029
Train on 186219 samples, validate on 20691 samples
Epoch 2/2
186219/186219 [==============================] - 84s 451us/step - loss: 0.4921 - acc: 
0.8071 - val_loss: 0.4617 - val_acc: 0.8168


verbose=2 (one line per epoch)

Train on 186219 samples, validate on 20691 samples
Epoch 1/1
 - 88s - loss: 0.5746 - acc: 0.7753 - val_loss: 0.4816 - val_acc: 0.8075
Train on 186219 samples, validate on 20691 samples
Epoch 1/1
 - 88s - loss: 0.4880 - acc: 0.8076 - val_loss: 0.5199 - val_acc: 0.8046
 
'''