Andrew Glassner - Deep Learning: A Visual Approach

Andrew Glassner

San Francisco

Deep Learning: A Visual Approach. Copyright 2021 by Andrew Glassner.

All rights reserved. No part of this work may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or by any information storage or retrieval system, without the prior written permission of the copyright owner and the publisher.

ISBN-13: 978-1-7185-0072-3 (print)
ISBN-13: 978-1-7185-0073-0 (ebook)

Publisher: William Pollock
Executive Editor: Barbara Yien
Production Editors: Maureen Forys and Rachel Monaghan
Developmental Editor: Alex Freed
Cover and Interior Design: Octopod Studios
Cover Illustrator: Gina Redman
Technical Reviewers: George Hosu and Ron Kneusel
Copyeditor: Rebecca Rider
Compositor: Maureen Forys, Happenstance Type-O-Rama
Proofreader: James Fraleigh

With the exception of the images noted at the end of the book in the Image Credits, all the images in this book are produced by the author. All original images may be freely downloaded from https://github.com/blueberrymusic and used as the reader pleases.

For information on book distributors or translations, please contact No Starch Press, Inc. directly:
No Starch Press, Inc.
245 8th Street, San Francisco, CA 94103
phone: 1-415-863-9900; info@nostarch.com
www.nostarch.com

Library of Congress Cataloging-in-Publication Data

Names: Glassner, Andrew S., author.

Title: Deep learning : a visual approach / Andrew Glassner.

Description: San Francisco, CA : No Starch Press, Inc., [2021] | Includes

bibliographical references and index.

Identifiers: LCCN 2020047326 (print) | LCCN 2020047327 (ebook) | ISBN

9781718500723 (paperback) | ISBN 9781718500730 (ebook)

Subjects: LCSH: Machine learning. | Neural networks.

Classification: LCC Q325.5 .G58 2021 (print) | LCC Q325.5 (ebook) | DDC

006.3/1--dc23

LC record available at https://lccn.loc.gov/2020047326

LC ebook record available at https://lccn.loc.gov/2020047327

No Starch Press and the No Starch Press logo are registered trademarks of No Starch Press, Inc. Other product and company names mentioned herein may be the trademarks of their respective owners. Rather than use a trademark symbol with every occurrence of a trademarked name, we are using the names only in an editorial fashion and to the benefit of the trademark owner, with no intention of infringement of the trademark.

The information in this book is distributed on an As Is basis, without warranty. While every precaution has been taken in the preparation of this work, neither the author nor No Starch Press, Inc. shall have any liability to any person or entity with respect to any loss or damage caused or alleged to be caused directly or indirectly by the information contained in it.

For Niko,
whos always there
with a smile,
a paw,
and a wag.

About the Author

Dr. Andrew Glassner is a Senior Research Scientist at Weta Digital, where he uses deep learning to help artists produce visual effects for film and television. He was Technical Papers Chair for SIGGRAPH 94, Founding Editor of the Journal of Computer Graphics Tools, and Editor-in-Chief of ACM Transactions on Graphics. His prior books include the Graphics Gems series and the textbook Principles of Digital Image Synthesis. Glassner holds a PhD from UNC-Chapel Hill. He paints, plays jazz piano, and writes novels. His website is www.glassner.com, and he can be followed on Twitter as @AndrewGlassner.

About the Technical Reviewers

George Hosu is a software engineer and world traveler with a broad interest in statistics and machine learning. He works as the lead machine learning engineer on an autoML and explainable AI project called Mindsdb. In his spare time, he writes to strengthen his understanding of epistemology, ML, and classical mathematics, and how they all come together to generate a meaningful map of the world. You can find his writing at https://blog.cerebralab.com/.

Ron Kneusel has been working with machine learning in industry since 2003 and completed a PhD in machine learning from the University of Colorado, Boulder, in 2016. Ron currently works for L3Harris Technologies, Inc. He has two books available from Springer: Numbers and Computers, and Random Numbers and Computers.

Acknowledgments

Authors like to say that nobody writes a book alone. We say that because its true. It gives me great pleasure to thank my friends and colleagues who helped me make this book possible.

For their consistent and enthusiastic support of this project, and for helping me feel good about it all the way through, I am enormously grateful to Eric Braun, Steven Drucker, Eric Haines, and Morgan McGuire. Thanks also to Georgia, Jenn, and Michael Ambrose for always providing cheerful conversation after Id spent too long at the computer.

Thanks to the reviewers of this books first edition, whose generous and insightful comments greatly improved the presentation: Adam Finkelstein, Alex Colburn, Alyn Rockwood, Angelo Pesce, Barbara Mones, Brian Wyvill, Craig Kaplan, Doug Roble, Eric Braun, Greg Turk, Jeff Hultquist, Kristi Morton, Lesley Istead, Matt Pharr, Mike Tyka, Morgan McGuire, Paul Beardsley, Paul Strauss, Peter Shirley, Philipp Slusallek, Serban Porumbescu, Stefanus Du Toit, Steven Drucker, Wenhao Yu, and Zackory Erickson.

Special thanks to super reviewers Alexander Keller, Eric Haines, Jessica Hodgins, and Luis Alvarado, who read the whole first edition and offered wonderful feedback on both presentation and structure. Thank you to technical reviewers George Hosu and Ron Kneusel for their insights.

Thanks to Todd Szymanski for advice on the design and layout of the books first edition and to Morgan McGuire for the Markdeep layout system, which enabled me to focus on writing and not word processing mechanics.

Thank you to the wonderful folks at No Starch Press for taking on this large project. Giant thanks to my editor Alex Freed, who read through the entire manuscript and offered numerous insightful comments and suggestions that greatly improved it throughout. Thank you to copyeditor Rebecca Rider and production editor Maureen Forys. Thank you to Rachel Monaghan for shepherding this project to completion with skill and grace. Thanks to publisher Bill Pollock for believing in the book and supporting the process.

My terrific colleagues at Weta Digital, Ltd. gave me professional and personal support and encouragement as I tackled this second edition. Thank you to Antoine Bouthors, Jedrzej Wojtowicz, Joe Letteri, Luca Fascione, Millie Maier, Navi Brouwer, Tom Buys, and Yann Provencher.

Introduction

Imagine that youre rubbing a golden lamp. You say, Genie, for my three wishes, give me someone to love, great wealth, and a long and healthy life.

Now imagine that youre entering your home. You say, House, bring the car around, ask Sarah if shes free for lunch, schedule a haircut, and make me a latte. Oh, and play some Thelonious Monk, please.

In both of these situations, youre asking a disembodied being of great power to hear you, understand you, and fulfill your desires. The first scenario is a fantasy going back millennia. The second scenario is commonplace reality today, thanks to artificial intelligence, or AI.