In [0]:
from __future__ import absolute_import, division, print_function, unicode_literals
# TensorFlow and tf.keras
# Have to install TensorFlow 2.0 with new session (but not each new runtime)
# !pip install tensorflow=="2.0.0"
import tensorflow as tf
from tensorflow import keras
#from keras.models import Sequential
#from keras.layers import Dense, Conv2D, Dropout, MaxPooling2D, Flatten
from keras.utils import np_utils
# Helper libraries
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import time
print(tf.__version__)
In [0]:
# Download data
(x_train, y_train), (x_test, y_test) = tf.keras.datasets.fashion_mnist.load_data()
class_names = ['T-shirt/top', 'Trouser', 'Pullover', 'Dress', 'Coat', 'Sandal', 'Shirt', 'Sneaker', 'Bag', 'Ankle boot']
In [0]:
plt.figure(figsize=(10,10))
for i in range(25):
plt.subplot(5,5,i+1)
plt.xticks([])
plt.yticks([])
plt.grid(False)
plt.imshow(x_train[i], cmap = plt.cm.binary)
plt.xlabel(class_names[y_train[i]])
plt.show()
In [0]:
# Data prep
# Normalize data
x_train = x_train / 255.0
x_test = x_test / 255.0
# add empty color dimension
x_train = np.expand_dims(x_train, -1)
x_test = np.expand_dims(x_test, -1)
In [0]:
Out[0]:
In [0]:
# TensorFlow/Keras Model
# https://colab.research.google.com/github/tensorflow/tpu/blob/master/tools/colab/fashion_mnist.ipynb?fbclid=IwAR070qdCr-MYoVJs0ovnv8nQCta5-GNzEfXgj_sXl31T53PPjRc5nGLh-w0#scrollTo=pWEYmd_hIWg8
def create_model():
model = tf.keras.models.Sequential()
model.add(tf.keras.layers.BatchNormalization(input_shape=x_train.shape[1:]))
model.add(tf.keras.layers.Conv2D(64, (5, 5), padding='same', activation='elu'))
model.add(tf.keras.layers.MaxPooling2D(pool_size=(2, 2), strides=(2,2)))
model.add(tf.keras.layers.Dropout(0.25))
model.add(tf.keras.layers.BatchNormalization(input_shape=x_train.shape[1:]))
model.add(tf.keras.layers.Conv2D(128, (5, 5), padding='same', activation='elu'))
model.add(tf.keras.layers.MaxPooling2D(pool_size=(2, 2)))
model.add(tf.keras.layers.Dropout(0.25))
model.add(tf.keras.layers.BatchNormalization(input_shape=x_train.shape[1:]))
model.add(tf.keras.layers.Conv2D(256, (5, 5), padding='same', activation='elu'))
model.add(tf.keras.layers.MaxPooling2D(pool_size=(2, 2), strides=(2,2)))
model.add(tf.keras.layers.Dropout(0.25))
model.add(tf.keras.layers.Flatten())
model.add(tf.keras.layers.Dense(256))
model.add(tf.keras.layers.Activation('elu'))
model.add(tf.keras.layers.Dropout(0.5))
model.add(tf.keras.layers.Dense(10))
model.add(tf.keras.layers.Activation('softmax'))
return model
In [0]:
# Initialize TPUs
import os
tf.config.experimental_connect_to_host('grpc://' + os.environ['COLAB_TPU_ADDR'])
resolver = tf.distribute.cluster_resolver.TPUClusterResolver('grpc://' + os.environ['COLAB_TPU_ADDR'])
tf.tpu.experimental.initialize_tpu_system(resolver)
strategy = tf.distribute.experimental.TPUStrategy(resolver)
tf.compat.v1.disable_eager_execution()
In [0]:
with strategy.scope():
keras_model = create_model()
keras_model.compile(
optimizer=tf.keras.optimizers.Adam(learning_rate=1e-3, ),
loss='sparse_categorical_crossentropy',
metrics=['sparse_categorical_accuracy'])
In [0]:
# Send model to TPUs
t1 = time.time()
keras_model.fit(
x_train.astype(np.float32), y_train.astype(np.float32),
epochs = 20,
steps_per_epoch = 1500,
# steps_per_epoch = 60, for some reason it crashes if I don't specify steps_per_epoch or if steps = nrow(train)/epochs
validation_data = (x_test.astype(np.float32), y_test.astype(np.float32)),
validation_freq = 1
)
t2 = time.time()
# We can save our model with: not sure where it is going
keras_model.save('model.h5')
# and reload it with:
# keras_model = tf.keras.models.load_model('model.h5')
train_time = t2 - t1
In [0]:
print(keras_model.summary())
print(train_time)
In [0]:
def training_curves(training, validation, title, subplot):
ax = plt.subplot(subplot)
ax.plot(training)
ax.plot(validation)
ax.set_title('model '+ title)
ax.set_ylabel(title)
ax.set_xlabel('epoch')
ax.legend(['training', 'validation'])
plt.subplots(figsize=(10,10))
plt.tight_layout()
training_curves(keras_model.history.history['sparse_categorical_accuracy'],
keras_model.history.history['val_sparse_categorical_accuracy'], 'accuracy', 211)
training_curves(keras_model.history.history['loss'],
keras_model.history.history['val_loss'], 'loss', 212)
In [0]:
# https://gist.github.com/yufengg/2b2fd4b81b72f0f9c7b710fa87077145
from sklearn import datasets, linear_model
from sklearn.metrics import accuracy_score
from sklearn.metrics import r2_score
from sklearn.metrics import f1_score
x_train_flat = np.array(x_train).reshape((-1, 28*28))
x_test_flat = np.array(x_test).reshape((-1, 28*28))
regr = linear_model.LinearRegression()
# Train the model using the training sets
regr.fit(x_train_flat, y_train)
Out[0]:
In [0]:
y_pred = regr.predict(x_test_flat).round(0)
In [0]:
accuracy_score(y_test, y_pred)
Out[0]:
In [0]:
r2_score(y_test, y_pred)
Out[0]:
In [0]:
y_pred.min()
Out[0]: