model.py

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import tensorflow
from tensorflow import keras
from keras.datasets import mnist
from keras.models import Sequential
from keras.layers import Dense, Dropout, Conv2D, MaxPooling2D, Flatten
from keras.callbacks import EarlyStopping
from keras import backend


# load dataset directly from keras library
(X_train, Y_train), (X_test, Y_test) = mnist.load_data()

# pre-process data
X_train = X_train.reshape((60000, 28, 28, 1)).astype('float32') / 255
X_test = X_test.reshape((10000, 28, 28, 1)).astype('float32') / 255

Y_train = keras.utils.to_categorical(Y_train)
Y_test = keras.utils.to_categorical(Y_test)

# define a CNN model
def create_model():
    num_classes = 10
    model = Sequential()
    model.add(Conv2D(32, (3, 3), activation='relu', input_shape=(28, 28, 1)))
    model.add(MaxPooling2D((2, 2)))
    model.add(Conv2D(64, (3, 3), activation='relu'))
    model.add(Dropout(0.25))
    model.add(Flatten())
    model.add(Dense(256, activation='relu'))
    model.add(Dropout(0.25))
    model.add(Dense(num_classes, activation='softmax'))
    model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])
    return model

# build the model
model = create_model()

# fit the model
print("Training the model . . .")
model.fit(X_train, Y_train, validation_data=(X_test, Y_test), epochs=10, batch_size=200, verbose=2)
print("The model has been successfully trained")

# save the model
print("Saving this model . . .")
model.save('model.h5')
print("The model has been successfully saved")

# evaluate the model
scores = model.evaluate(X_test, Y_test, verbose=0)
print("Neural network error: %.2f%%" % (100-scores[1]*100))