The Master Algorithm: How the Quest for the Ultimate Learning Machine Will Remake Our World
The Master Algorithm: How the Quest for the Ultimate Learning Machine Will Remake Our World book cover

The Master Algorithm: How the Quest for the Ultimate Learning Machine Will Remake Our World

by
Price
$10.99
Format
Paperback
Pages
352
Publisher
Basic Books
Publication Date
ISBN-13
978-0465094271
Dimensions
5.45 x 1.2 x 8.25 inches
Weight
10.2 ounces

Description

"Wonderfully erudite, humorous, and easy to read."― KDNuggets "Pedro Domingos demystifies machine learning and shows how wondrous and exciting the future will be."― Walter Isaacson, New York Times bestselling author of Steve Jobs, The Innovators, and The Code Breaker "An impressive and wide-ranging work that covers everything from the history of machine learning to the latest technical advances in the field."― Daily Beast "Domingos writes with verve and passion."― New Scientist "Unlike other books that proclaim a bright future, this one actually gves you what you need to understand the changes that are coming."― Peter Norvig, Director of Research, Google andcoauthor of Artificial Intelligence: A Modern Approach "Domingos is the perfect tour guide from whom you will learn everything you need to know about this exciting field, and a surprising amount about sience and philosophy as well."― Duncan Watts, Principal Researcher, Microsoft Research, and author of Six Degrees and Everything Is Obvious *Once You Know the Answer "[ The Master Algorithm ] does a good job of examining the field's five main techniques.... The subject is meaty and the author...has a knack for introducing concepts at the right moment."― The Economist "Domingos is a genial and amusing guide, who sneaks us around the backstage areas of the science in order to witness the sometimes personal (and occasionally acrimonious) tenor of research on the subject in recent decades."― Times Higher Education "An exhilarating venture into groundbreaking computer science." ― Booklist, starred review "[An] enthusiastic but not dumbed-down introduction to machine learning...lucid and consistently informative.... With wit, vision, and scholarship, Domingos decribes how these scientists are creating programs that allow a computer to teach itself. Readers...will discover fascinating insights." ― Kirkus Reviews "This book is a must have to learn machine learning without equation. It will help you get the big picture of the several learning paradigms. Finally, the provocative idea is not only intriguing, but also very well argued."― Data Mining Research "If you are interested in a crash course on the enigmatic field of machine learning and the challenges for AI practitioners that lie ahead, this book is a great read."― TechCast Global "This book is a sheer pleasure, mixed with education. I am recommending it to all my students, those who studied machine learning, those who are about to do it and those who are about to teach it."― Judea Pearl, author of The Book of Why, and professor of computer science, UCLA and winner of the A. M. Turing Award "Machine learning is the single most transformative technology that will shape our lives over the next fifteen years. This book is a must-read--a bold and beautifully written new framework for looking into the future."― Geoffrey Moore, author of Crossing the Chasm "Machine learning is a fascinating world never before glimpsed by outsiders. Pedro Domingos initiates you to the mysterious languages spoken by its five tribes, and invites you to join in his plan to unite them, creating the most powerful technology our civilization has ever seen."― Sebastian Seung, professor, Princeton, and author of Connectome "A delightful book by one of the leading experts in the field. If you wonder how AI will change your life, read this book."― Sebastian Thrun, research professor, Stanford, Google Fellow and Inventor of the Self-Driving Car "This is an incredibly important and useful book. Machine learning is already critical to your life and work, and will only become more so. Finally, Pedro Domingos has written about it in a clear and understandable fashion."― Thomas H. Davenport, Distinguished Professor, Babson College and author of Competing on Analytics and Big Data @ Work span "Machine learning, known in commercial use as predictive analytics, is changing the world. This riveting, far-reaching, and inspiring book introduces the deep scientific concepts to even non-technical readers, and yet also satisfies experts with a fresh, profound perspective that reveals the most promising research directions. It's a rare gem indeed."― Eric Siegel, founder of Predictive Analytics World and author of Predictive Analytics: The Power to Predict Who Will Click, Buy, Lie, or Die Pedro Domingos is a professor emeritus of computer science at the University of Washington. He is a winner of the SIGKDD Innovation Award, the highest honor in data science. A fellow of the Association for the Advancement of Artificial Intelligence, he lives near Seattle.

Features & Highlights

  • Recommended by Bill GatesA thought-provoking and wide-ranging exploration of machine learning and the race to build computer intelligences as flexible as our own
  • In the world's top research labs and universities, the race is on to invent the ultimate learning algorithm: one capable of discovering any knowledge from data, and doing anything we want, before we even ask. In
  • The Master Algorithm
  • , Pedro Domingos lifts the veil to give us a peek inside the learning machines that power Google, Amazon, and your smartphone. He assembles a blueprint for the future universal learner--the Master Algorithm--and discusses what it will mean for business, science, and society. If data-ism is today's philosophy, this book is its bible.

Customer Reviews

Rating Breakdown

★★★★★
30%
(488)
★★★★
25%
(407)
★★★
15%
(244)
★★
7%
(114)
★
23%
(375)

Most Helpful Reviews

✓ Verified Purchase

This book is not for complete noobs/beginners

This is a great book and covers everything but this book is certainly not for someone without prior knowledge of machine learning and good understand of some principles.

At the start of the book, Pedro mentions that this book is written for the layman but it's not.

I gave this book to a friend of mine before I read it and he was like "doode this is not for laymen, I didn't understand a thing".

While the book is great for seasoned programmers and those starting with machine learning and good knowledge of some concepts, this book ain't for someone who wants to know more about machine learning and algorithms.

I still give it 5 because Pedro is a great guy and this book is indepth.
12 people found this helpful
✓ Verified Purchase

Skip Chapters 1-3 and 9-10

Chapters four through eight were valuable to me as an introduction to machine learning and the five different approaches described by the author. The first three chapters and final two seems to be wishful thinking on the part of the author. The question I kept asking my self was, "if there is a master algorithm and the existing approaches are spanned from it, why hasn't the author been able to trace the branches back to the trunk and found the grand prize? The way he describes it, they should be back traceable, simply, and elegantly.
9 people found this helpful
✓ Verified Purchase

Mediocre book at best

I recently received a Computer Science degree from UCLA. This book is so outdated in its description of what would be considered current AI scholarship, it's surprising it was only written not long ago. It's an interesting read if you're looking for a historical book.
2 people found this helpful
✓ Verified Purchase

OK for beginners or the layperson, otherwise don't waste your time

I was skeptical of the book early on, but decided to withhold judgment and finish it before deciding. Unfortunately, I wasn’t rewarded for the effort. My initial feelings were stirred by the pejorative way Dr. Domingos refers to his peers and colleagues—the “five tribes” of machine learning. Regardless if his categorization is correct or not, the language makes it clear he is looking down upon them as a more primitive group in the field in which he feels dominant. It took a bit longer to grasp what the title meant—the idea of a “Master Algorithm” is less a high abstraction than a statement of Dr. Domingos’ vision. He believes he has seen the real truth, and can even discern its form, although at last admitting he doesn't have it. That last admission takes 292 pages.

Particularly grating to me, and ever more so as the narrative wore on, is the constant use of weak analogies and anthropomorphism in an attempt to make the subject matter more attainable to us mere mortals. Towards the end of the book we are subjected to a weak fantasy of a 5-sided magical land in the center of which rises a testament to the beauty of the Master Algorithm. Before that, we suffer with the story of Robby the Robot, who starts as a child, and we must figure out not how to teach him (it?) but how to get him to teach himself by learning. The subject material is important, even critical to the overall picture, but these distractions make the topic less, rather than more, understandable. As well, the story is punctuated here and there with odd personal remarks like “looks like a cross between …”.

We are never far either from reminders that Dr. Domingos is a font of Machine Learning discovery and knowledge, to wit, in reference to Naïve Bayes learners “when I belatedly decided to include Naïve Bayes in my experiments, I was shocked to find it did better than all the other the algorithms I was comparing, save one—luckily, the algorithm I was developing for my thesis”. (Emphasis added.) Or: “RISE, as I called the algorithm…made better predictions than the best rule-based and instance-based learners of the time”. Or: (in reference to modeling word of mouth and influence) “Can we measure each member’s influence and target just enough of the most influential members…With my student Matt Richardson, I designed an algorithm that did just that…An avalanche of other research on this problem followed.” OK, I get it, Dr. Dominos. You have done a lot of work in this field, including your share of invention, and you really know your stuff—but that isn't what the book is about, is it?

Early in the book and here and there throughout, Dr. Domingos puts a lot of weight behind machine learning as the way forward to cure cancer, down to individual drug design and treatment (a hot topic these days). I feel this and other examples are too much of a stretch—it’s as if the author shifts from a reporter to a marketer for machine learning at many points. Later in the book I think Dr. Domingos goes way too far. First, he confounds the idea of robotics with the idea of machine learning and so-called Artificial Intelligence. He then veers into philosophy, telling us that robot war will be good for us, then spending some time trying to convince us not to worry about rogue robot intelligence. His arguments for the “don’t worry, be happy” position are mainly built on the idea that robots can only do what we tell them. Unfortunately he points out that future AIs might “give us what we ask for instead of what we want”.

The book is very useful to understand there are many flavors of problems and machine learning, and to consider the idea that the best future learners will be combinations of some or all of them. While it is written to be very approachable, for the more technology or science minded I think it goes too far. However, there is a large audience who, to be fair, that Dr. Domingos points out needs to know about these topics yet are not scientists or programmers. For that audience I think the book does a good service, but if you are more interested in the science and technology of Machine Learning, I would look elsewhere.
2 people found this helpful
✓ Verified Purchase

Outdated

This book is very outdated and no longer relevant. You will notice that even the title uses outdated words that serious computer scientists have chosen to move away from. The author seems strangely proud to be holding on to computing norms no longer accepted by the rest of the community, which is odd given how the rest of the field tries to use its accumulated knowledge to advance itself.
1 people found this helpful
✓ Verified Purchase

Great overview but somewhat disjointed

This book provides a lot of details on machine learning and introduced me to many new terms and ideas. The way machine learning is described is that the typical method for computing is to give the machine an input and write the algorithm and then receive the output, whereas with machine learning you give the computer inputs and outputs and it writes the algorithm.

The main theme is to integrate all of the current methods to form a master algorithm that can do everything. While this sounds daunting, the author builds up his case starting from describing the current (as of 2015) five main approaches: symoblists, connectionists, evolutionaries, bayesians, and analogizers. Each group is broken down into their formal language used, method of evaluation, and method to evaluate and refine. This history is enough to recommend the book and Domingos does a good job of comparing and contrasting the different tribes. He coves the ubiquitous problem of over-fitting and explains why testing on unseen data is the ultimate verification method.

While there are many interesting ideas, I found a lot of the presentation to be unclear and somewhat vague. This could be due to my ignorance of the subject. He also undercuts his entire thesis at the beginning by granting Hume's problem of induction as a given. This demonstrates the importance of the proper philosophical ideas when addressing fundamental topics.

I am actually less optimistic about all of this after reading this. It still seems very far off with some major obstacles. With that being said, I admire the author's passion and interest in the subject and am grateful to have the information he's provided. I noticed he now works in the finance industry and I would be very interested to hear if he has more or less confidence with Machine Learning now versus when the book was written, over 5 years ago. Even with its faults, highly recommend.
1 people found this helpful
✓ Verified Purchase

Great overview but somewhat disjointed

This book provides a lot of details on machine learning and introduced me to many new terms and ideas. The way machine learning is described is that the typical method for computing is to give the machine an input and write the algorithm and then receive the output, whereas with machine learning you give the computer inputs and outputs and it writes the algorithm.

The main theme is to integrate all of the current methods to form a master algorithm that can do everything. While this sounds daunting, the author builds up his case starting from describing the current (as of 2015) five main approaches: symoblists, connectionists, evolutionaries, bayesians, and analogizers. Each group is broken down into their formal language used, method of evaluation, and method to evaluate and refine. This history is enough to recommend the book and Domingos does a good job of comparing and contrasting the different tribes. He coves the ubiquitous problem of over-fitting and explains why testing on unseen data is the ultimate verification method.

While there are many interesting ideas, I found a lot of the presentation to be unclear and somewhat vague. This could be due to my ignorance of the subject. He also undercuts his entire thesis at the beginning by granting Hume's problem of induction as a given. This demonstrates the importance of the proper philosophical ideas when addressing fundamental topics.

I am actually less optimistic about all of this after reading this. It still seems very far off with some major obstacles. With that being said, I admire the author's passion and interest in the subject and am grateful to have the information he's provided. I noticed he now works in the finance industry and I would be very interested to hear if he has more or less confidence with Machine Learning now versus when the book was written, over 5 years ago. Even with its faults, highly recommend.
1 people found this helpful
✓ Verified Purchase

Outdated and too technical

I wouldn't recommend this book to anyone. If you don't know anything about ML then this book will be too technical and if you do then you wont get much out of it. The historical parts of the book might be the only reason I'd recommend it.
1 people found this helpful
✓ Verified Purchase

Very limited scope

Make no mistake this is a propaganda book, it was highly touted in the office by managers so our engineering team starting ML took up this book, thus wasting everyone's time. This is the car salesman of ML books, it tells you just enough to hopefully interest you in the topic.

So this is a perfectly fine book if you are trying to propagandize your manager to let you do some ML work, but that is its scope. Don't expect to use it for learning about ML, implementation, functional or technical details... or even a high-level analogy.
1 people found this helpful
✓ Verified Purchase

The best time to commit a burglary is during an earthquake

Mostly I learned that the best time to commit a burglary is during an earthquake.
1 people found this helpful

Related Books