In [0]:
from matplotlib import pyplot as plt
import numpy as np
from imageio import imread
import pandas as pd
from time import time as timer
import tensorflow as tf
%matplotlib inline
from matplotlib import animation
from IPython.display import HTML
EXERCISE 1: Train deeper network¶
Make a deeper model, with wider layers. Remember to 'softmax' activation in the last layer, as required for the classification task to encode pseudoprobabilities. In the other layers you could use 'relu'.
Try to achieve 90% accuracy. Does your model overfit?
In [0]:
fashion_mnist = tf.keras.datasets.fashion_mnist
(x_train, y_train), (x_test, y_test) = fashion_mnist.load_data()
x_train = x_train/255
x_test = x_test/255
class_names = ['T-shirt/top', 'Trouser', 'Pullover', 'Dress', 'Coat',
'Sandal', 'Shirt', 'Sneaker', 'Bag', 'Ankle boot']
In [3]:
# 1. create model
model = tf.keras.models.Sequential([
tf.keras.layers.Flatten(input_shape=(28, 28)),
tf.keras.layers.Dense(1024, activation='relu'),
tf.keras.layers.Dense(256, activation='relu'),
tf.keras.layers.Dense(64, activation='relu'),
tf.keras.layers.Dense(10, activation='softmax')
])
model.compile(optimizer='adam',
loss='sparse_categorical_crossentropy',
metrics=['accuracy'])
model.summary()
# 2. train the model
save_path = 'save/mnist_{epoch}.ckpt'
save_callback = tf.keras.callbacks.ModelCheckpoint(filepath=save_path, save_weights_only=True)
hist = model.fit(x=x_train, y=y_train,
epochs=20, batch_size=128,
validation_data=(x_test, y_test),
callbacks=[save_callback])
# 3. plot the loss and accuracy evolution during training
fig, axs = plt.subplots(1, 2, figsize=(10,5))
axs[0].plot(hist.epoch, hist.history['loss'])
axs[0].plot(hist.epoch, hist.history['val_loss'])
axs[0].legend(('training loss', 'validation loss'), loc='lower right')
axs[1].plot(hist.epoch, hist.history['accuracy'])
axs[1].plot(hist.epoch, hist.history['val_accuracy'])
axs[1].legend(('training accuracy', 'validation accuracy'), loc='lower right')
plt.show()
# 4. evaluate model in best point (before overfitting)
model.load_weights('save/mnist_10.ckpt')
model.evaluate(x_test, y_test, verbose=2)
Out[3]: