Neural Network Projects with Python: The ultimate guide to using Python to explore the true power of neural networks through six projects

301 Pages·2019·8.139 MB·English

by  James Loy

Neural Network Projects
with Python
The ultimate guide to using Python to explore the true power
of neural networks through six projects
James Loy
BIRMINGHAM - MUMBAI
Neural Network Projects with Python
Copyright © 2019 Packt Publishing
All rights reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted in any form
or by any means, without the prior written permission of the publisher, except in the case of brief quotations
embedded in critical articles or reviews.
Every effort has been made in the preparation of this book to ensure the accuracy of the information presented.
However, the information contained in this book is sold without warranty, either express or implied. Neither the
author, nor Packt Publishing or its dealers and distributors, will be held liable for any damages caused or alleged to
have been caused directly or indirectly by this book.
Packt Publishing has endeavored to provide trademark information about all of the companies and products
mentioned in this book by the appropriate use of capitals. However, Packt Publishing cannot guarantee the accuracy
of this information.
Commissioning Editor: Pravin Dhandre
Acquisition Editor: Nelson Morris
Content Development Editor: Pratik Andrade
Technical Editor: Jovita Alva
Copy Editor: Safis Editing
Project Coordinator: Namrata Swetta
Proofreader: Safis Editing
Indexer: Rekha Nair
Graphics: Jisha Chirayil
Production Coordinator: Arvindkumar Gupta
First published: February 2019
Production reference: 1270219
Published by Packt Publishing Ltd.
Livery Place
35 Livery Street
Birmingham
B3 2PB, UK.
ISBN 978-1-78913-890-0
www.packtpub.com
To my wife Agnes Lim - my partner, soulmate, and cheerleader.
Without her, this book would not have been possible.
- James Loy
mapt.io
Mapt is an online digital library that gives you full access to over 5,000 books and videos, as
well as industry leading tools to help you plan your personal development and advance
your career. For more information, please visit our website.
Why subscribe?
Spend less time learning and more time coding with practical eBooks and Videos
from over 4,000 industry professionals
Improve your learning with Skill Plans built especially for you
Get a free eBook or video every month
Mapt is fully searchable
Copy and paste, print, and bookmark content
Packt.com
Did you know that Packt offers eBook versions of every book published, with PDF and
ePub files available? You can upgrade to the eBook version at www.packt.com and as a print
book customer, you are entitled to a discount on the eBook copy. Get in touch with us at
[email protected] for more details.
At www.packt.com, you can also read a collection of free technical articles, sign up for a
range of free newsletters, and receive exclusive discounts and offers on Packt books and
eBooks.
Contributors
About the author
James Loy has more than five years, expert experience in data science in the finance and
healthcare industries. He has worked with the largest bank in Singapore to drive
innovation and improve customer loyalty through predictive analytics. He has also
experience in the healthcare sector, where he applied data analytics to improve decision-
making in hospitals. He has a master's degree in computer science from Georgia Tech, with
a specialization in machine learning.
His research interest includes deep learning and applied machine learning, as well as
developing computer-vision-based AI agents for automation in industry. He writes on
Towards Data Science, a popular machine learning website with more than 3 million views
per month.
About the reviewer
Mike Thompson has worked as a software engineer in a variety of roles involving data
engineering, service development, and distributed systems design over the past 8 years.
Mike worked for Bungie and helped ship game titles in the Destiny franchise. His credits
include Destiny in 2014 and Destiny 2 in 2017, including many expansions in between for
both titles. His contributions were integral to scaling Bungie's backend data infrastructure
and allowing millions of players to enjoy the Destiny games.
Mike now works for ProbablyMonsters, a growing family of game studios, where he
focuses on data services and infrastructure, along with many other duties that are required
to get a new business off the ground.
Packt is searching for authors like you
If you're interested in becoming an author for Packt, please visit authors.packtpub.com
and apply today. We have worked with thousands of developers and tech professionals,
just like you, to help them share their insight with the global tech community. You can
make a general application, apply for a specific hot topic that we are recruiting an author
for, or submit your own idea.
Table of Contents
Preface 1
Chapter 1: Machine Learning and Neural Networks 101 6
What is machine learning? 7
Machine learning algorithms 8
The machine learning workflow 10
Setting up your computer for machine learning 12
Neural networks 14
Why neural networks? 15
The basic architecture of neural networks 16
Training a neural network from scratch in Python 17
Feedforward 18
The loss function 19
Backpropagation 19
Putting it all together 21
Deep learning and neural networks 22
pandas – a powerful data analysis toolkit in Python 24
pandas DataFrames 24
Data visualization in pandas 27
Data preprocessing in pandas 30
Encoding categorical variables 31
Imputing missing values 33
Using pandas in neural network projects 35
TensorFlow and Keras – open source deep learning libraries 35
The fundamental building blocks in Keras 36
Layers – the atom of neural networks in Keras 36
Models – a collection of layers 37
Loss function – error metric for neural network training 37
Optimizers – training algorithm for neural networks 38
Creating neural networks in Keras 38
Other Python libraries 41
Summary 41
Chapter 2: Predicting Diabetes with Multilayer Perceptrons 42
Technical requirements 43
Diabetes – understanding the problem 44
AI in healthcare 45
Automated diagnosis 46
The diabetes mellitus dataset 47
Exploratory data analysis 47
Data preprocessing 54
Table of Contents
Handling missing values 54
Data standardization 58
Splitting the data into training, testing, and validation sets 59
MLPs 61
Model architecture 61
Input layer 62
Hidden layers 62
Activation functions 62
ReLU 63
Sigmoid activation function 64
Model building in Python using Keras 65
Model building 65
Model compilation 66
Model training 66
Results analysis 68
Testing accuracy 68
Confusion matrix 69
ROC curve 71
Further improvements 73
Summary 74
Questions 74
Chapter 3: Predicting Taxi Fares with Deep Feedforward Networks 77
Technical requirements 78
Predicting taxi fares in New York City 79
The NYC taxi fares dataset 80
Exploratory data analysis 80
Visualizing geolocation data 81
Ridership by day and hour 86
Data preprocessing 88
Handling missing values and data anomalies 88
Feature engineering 94
Temporal features 94
Geolocation features 95
Feature scaling 99
Deep feedforward networks 100
Model architecture 100
Loss functions for regression problems 101
Model building in Python using Keras 102
Results analysis 104
Putting it all together 108
Summary 111
Questions 111
Chapter 4: Cats Versus Dogs - Image Classification Using CNNs 113
[ ii ]
Table of Contents
Technical requirements 114
Computer vision and object recognition 115
Types of object recognition tasks 116
Digital images as neural network input 118
Building blocks of CNNs 120
Filtering and convolution 120
Max pooling 124
Basic architecture of CNNs 125
A review of modern CNNs 126
LeNet (1998) 126
AlexNet (2012) 126
VGG16 (2014) 127
Inception (2014) 127
ResNet (2015) 127
Where we stand today 128
The cats and dogs dataset 128
Managing image data for Keras 130
Image augmentation 132
Model building 134
Building a simple CNN 135
Leveraging on pre-trained models using transfer learning 140
Results analysis 143
Summary 147
Questions 148
Chapter 5: Removing Noise from Images Using Autoencoders 150
Technical requirements 150
What are autoencoders? 152
Latent representation 153
Autoencoders for data compression 155
The MNIST handwritten digits dataset 155
Building a simple autoencoder 157
Building autoencoders in Keras 158
Effect of hidden layer size on autoencoder performance 162
Denoising autoencoders 165
Deep convolutional denoising autoencoder 168
Denoising documents with autoencoders 172
Basic convolutional autoencoder 176
Deep convolutional autoencoder 179
Summary 183
Questions 184
Chapter 6: Sentiment Analysis of Movie Reviews Using LSTM 185
Technical requirements 186
[ iii ]