Machine Learning

The MIT Press Essential Knowledge

by Ethem Alpaydin

A concise overview of machine learning—computer programs that learn from data—which underlies applications that include recommendation systems, face recognition, and driverless cars. Today, machine learning underlies a range of applications we use every day, from product recommendations to voice recognition—as well as some we don't yet use everyday, including driverless cars. It is the basis of the new approach in computing where we do not write programs but collect data; the idea is to learn the algorithms for the tasks automatically from data. As computing devices grow more ubiquitous, a larger part of our lives and work is recorded digitally, and as “Big Data” has gotten bigger, the theory of machine learning—the foundation of efforts to process that data into knowledge—has also advanced. In this book, machine learning expert Ethem Alpaydin offers a concise overview of the subject for the general reader, describing its evolution, explaining important learning algorithms, and presenting example applications. Alpaydin offers an account of how digital technology advanced from number-crunching mainframes to mobile devices, putting today's machine learning boom in context. He describes the basics of machine learning and some applications; the use of machine learning algorithms for pattern recognition; artificial neural networks inspired by the human brain; algorithms that learn associations between instances, with such applications as customer segmentation and learning recommendations; and reinforcement learning, when an autonomous agent learns act so as to maximize reward and minimize penalty. Alpaydin then considers some future directions for machine learning and the new field of “data science,” and discusses the ethical and legal implications for data privacy and security.

Genres: Nonfiction, Computer Science, Artificial Intelligence, Science, Technology, Programming, Computers

232 pages, Kindle Edition

First published January 1, 2016

About the author

Ethem ALPAYDIN received his BSc from Department of Computer Engineering of Bogazici University in 1987 and the degree of Docteur es Sciences from Ecole Polytechnique Fédérale de Lausanne in 1990. He did his postdoctoral work at the International Computer Science Institute, Berkeley in 1991 and afterwards was appointed Assistant Professor at the Department of Computer Engineering of Bogazici University. He was appointed Associate Professor in 1996 and Professor in 2002 in the same department.
As visiting researcher, he worked at Department of Brain and Cognitive Sciences, MIT in 1994, International Computer Science Institute, Berkeley in 1997, IDIAP, Switzerland in 1998, and TU Delft in 2014.

He was Fulbright Senior Scholar in 1997/1998 and received the Research Excellence Award from the Bogazici University Foundation in 1998 (junior faculty) and 2008 (senior faculty), the Young Scientist Award from the Turkish Academy of Sciences in 2001 and the Scientific Encouragement Award from the Turkish Scientific and Technical Research Council in 2002.

His book Introduction to Machine Learning was published by The MIT Press in October 2004; its German edition was published by Oldenbourg Verlag in May 2008, and Chinese edition was published by Huazhang Press in June 2009. Introduction to Machine Learning, second edition was published by The MIT Press in February 2010; its Turkish edition was published by Bogazici University Press in April 2011 and Chinese edition was published by Huazhang Press in June 2014. Introduction to Machine Learning, third edition was published by The MIT Press in August 2014.

He was an Editorial Board Member of The Computer Journal (Oxford University Press) in 2008-2014. He is a Member of The Science Academy, Turkey, Senior Member of the IEEE, and an Editorial Board Member of Pattern Recognition (Elsevier).

Reviews

[Maryam Taheri · Rating 5 out of 5]

یک مرجع عالی برای یادگیری ماشین که از صفر تا صد همه‌‌ی بخش‌هاش رو با بیان خیلی خیلی ساده گفته و تمام چیزی که توی هر بخش لازمه رو‌میگه و هیچ‌چیز رو از قلم نمیندازه.

[Anders Brabaek · Rating 4 out of 5]

Summary: this book is for understanding the concepts of machine learning, not the doing, not the technology, and not the business it will drive. That is, it explains the math and statistics at a conceptual level where anyone can follow.

This book, oddly, starts by explaining the absolutely most trivial things about technology and the Internet – e.g. we now have smart phones. But once that part has past, the author Alpaydin explains the conceptual ideas behind the algorithms and the thinking surrounding Machine Learning, AI and neural networks. Alpaydin does this without ever becoming really technical, and this book is for understanding the basic concepts, not the doing.
At the very end, Alpaydin delivers a few good examples of some of the things which is currently in the making regard machine learning.
The book never touches on how you yourself, or your business can start playing with machine learning. Even so, by understanding the conceptual parts of machine learning, I believe many will have an intuitive idea about what can be in the making.

I give this book a rating 4 mostly because I believe it delivers what I expected in a decent well written way.

However, it contained no surprises (except for the extreme simple non fitting trivia in the beginning), and no exceptional insights – which I can promise do exist in this field. As for information about what sort of things machine learning will bring us, something like “Automate this” would be a better book (despite being dated, and only describing existing solutions).

But of course, for the doers, going to fx. the Microsoft Azure Machine learning websites or the AWS Machine learning websites would be a lot more beneficial. Books are not really a good source once the you start to dig into an area moving at this pace – well, with the exception of understanding the math, and the statistics.

[Zac · Rating 4 out of 5]

Recommended to me by a product manager at Hulu. It's not too technical, but I wish the book was condensed into more of a primer with more theory/conceptual discussion and examples rather than over-explaining technical details. The engineers reading will be sick of hearing it and the managers/non-engineers won't fully understand.

Overall, if you want to understand and introduction to machine learning and how it works, this book will do the job.

[Galina · Rating 4 out of 5]

As a person with no previous knowledge/experience in machine learning, I found this book very enlightening. It explains fairly well the general concepts in machine learning like classification, regression, clustering, reinforcement learning with clear examples without going too technical. It actually helped me to gain enough understanding to form better questions about machine learning to search for online, and understand the research I found, and generally got me very excited about the applications of machine learning and the technology that allows for performing it today. Exciting times we live in!

[Onuralp · Rating 2 out of 5]

Hardly qualify Essential Knowledge, better to read Wikipedia.

[meowdeleine · Rating 2 out of 5]

frankly i struggle to understand the appeal of “non-technical” explanations for "technical" topics .. i’d far prefer one good honest equation to all these summaries and summaries of summaries. but perhaps i’m being too critical? the task here is enormous. i understand how this book could be helpful for many people. but if my brother asked if he should read it, i'd tell him to just watch statquest on youtube and follow a few random chatty engineers on tech twitter instead

[Richard Thompson · Rating 3 out of 5]

I already knew a bit about machine learning from reading books written for a broad popular audience. This was the next step up. I learned a few new things about techniques for attacking different types of large data sets and for building neural networks, but a most of this won't stick and isn't in my mind in a way that I can usefully apply it. I doubt that I will ever become a serious AI researcher, but I'd like to be able to look under the hood enough to understand what others are doing and to be able to try some simple things myself so that I can develop ideas about new and interesting things that can be done with current and developing technology. To get there, I think that I need to start with a textbook and some opensource AI tools that I can play with so that I can build a few things and watch them go to work.

Any suggestions from my fellow Goodreaders as to the best way(s) to do this would be appreciated.

This was my first book in the MIT Essential Knowledge series. Some of the others also look interesting so it won't be my last.

[Andy · Rating 4 out of 5]

Pretty good introduction to the conceptual side of machine learning, for someone who knows a bit of programming and a bit of statistics. Skimmed over some of the beginning parts and some of the later parts went over my head. Fair warning: this will not teach you how to actually write scripts/programs, but it did inspire some improvements to my own (rudimentary) work.

My main criticism is that the author can trend towards the naïve (or idealistic? or complacent? or?) in the ending chapter about the risks of machine learning applications. (Most, if not all, techies lack moral compasses and it would be better to recognize that up front. Also, automated/self-driving personal cars are inherently violent. Sorry.)

[Nour Ibrahim · Rating 3 out of 5]

I think I wanted a nice introduction to machine learning and this was not it. I feel like the guy tried to explore so many things at once that if someone who's a total beginner like me won't feel satisfied and will actually feel depressed of how so many things didn't make sense. Still there were parts that I understood and I liked that his language wasn't always complicated. But I wouldn't recommend reading this on its own without already taking a course to actually understand the context and enjoy the whole book.

[Colin Thomson · Rating 3 out of 5]

This was an interesting and broad introduction to the topic of machine learning. It was accessible enough not to feel overwhelmed, and managed not to bombard you with technical terms. I read this for professional development, and feel I reaped the necessary benefits.
Given I normally read for pleasure, I'm not sure how to rate this. Maybe it's 4 stars?!?!
Regardless, I'd recommend it to anyone in tech.

[Jay French · Rating 3 out of 5]

I found this a good re-introduction to machine learning. By re-introduction, I’m using my perspective based on experience, having worked in AI and neural nets twenty years back, but keeping up through pop science magazine articles and such since then. For me, I was reminded of many of the methods I knew, and a few I hadn’t heard of. Short. Nice description, just what you want.

[Mark Higley · Rating 2 out of 5]

I had to force myself to finish it. But that reflects more on my ADD than the content of the book. It is a well written summary of the concept of machine learning and AI. If you're into computers and the future of technology, give it a read.

[Abdul · Rating 3 out of 5]

“Intelligence seems not to originate from some outlandish formula, but rather from the patient, almost brute force use of simple, straightforward algorithms. It"

My Take:
Machine Learning and AI are a fact. They have arrived. It's not a new phenomenon or technology but something that always existed but they are now for the first time, more accessible to businesses and even the average person without realizing.

Did you know that, for example, Judges ins some countries rely partly on a report that has been created using machine learning and artificial intelligence about the likelihood of someone re-offending before they decide on a parole hearing whether it is safe to release an inmate or not?

I have read enough articles and news about Machine Learning and AI but had no full picture of the technology and I have always been curious about this subject which will become more prominent in the future. So I have decided to read an introductory book about Machine Learning and AI.

This book gives a full overview of the subject without diving deep into any aspect of it. It is a very short book and although it is meant for anyone, despite the best effort of the author who intended it to be a non-geek's intro to Machine Learning and AI, it is sometimes discussing aspects that I wonder if the non-computer scientist will appreciate fully.

However, this does not take away the main goal of the book which is to be accessible to the non-computer scientist. If you struggle to understand a certain concept or a phrase, you can safely ignore it and carry on reading. It won't spoil the book for you and you won't miss it I promise. It was most likely a deeper look at a concept and you only need a 10,000-foot view.

All in all a good book although a little bit on the dry and scientific side. I would have given it more stars if it was more entertaining and had more real-world examples for the non-computer scientists.

Quotes:

“Each of us, actually every animal, is a data scientist. We collect data from our sensors, and then we process the data to get abstract rules to perceive our environment and control our actions in that environment to minimize pain and/or maximize pleasure. We have memory to store those rules in our brains, and then we recall and use them when needed. Learning is lifelong; we forget rules when they no longer apply or revise them when the environment changes. Learning"

“After all, this is what learning is: as we learn a task, we get better at it, be it tennis, geometry, or a foreign language"

“What we lack in knowledge, we make up for in data. "

“Now data is cheap and we are all kings and queens of our little online fiefdoms. A baby born to gadget-loving parents today can generate more data in her first month than it took for Homer to narrate the complete adventures of Odysseus."

“With social media, each of us is now a celebrity whose life is worth following. We are our own paparazzi. We are no longer only allotted 15 minutes of fame, but every time we are online we are famous. The social media allows us to write our digital autobiography as we are living it."

“With the number of smart machines continuously helping us in our daily lives, we all become producers of data. Every time we buy a product, rent a movie, visit a web page write a blog or post on social media, even when we walk or drive around, we generate data, and that data is valuable for someone who is interested in collecting and analyzing it. The customer is not only always right, but also interesting and worth following."

[Stefan Kanev · Rating 3 out of 5]

This is a curious book.

I'm mostly grateful that it introduced me to the MIT Essential Knowledge series, which seem pretty promising. They are short, small and beautifully produced. I ordered a bunch of others and am quite keen to start reading them.

That being said, this book had a lot of promise, but I don't feel it delivered fully. It's a very high-level overview of what Machine Learning is and not much more. It doesn't go to code (or even math), but uses well-written prose to explain each key idea that you will encounter in a detail ML course. It's not incredibly well-structured, and it does not attempt to present the material in a good hierarchical way. You have to read carefully and pay a lot of attention, otherwise it's easy to get lost. The author shares some high-level thoughts about automation and where the world is going with it, some of which very insightful, some of which a bit out there.

That being said, I would still recommend it. It's mostly a "non-technical CEO" book – if you want to get a high-level sense of what ML is, without going in the weeds, it's worth checking out. Even better if you like beautiful books.

[Garry Cotton · Rating 5 out of 5]

Excellent book, does a fantastic of job of balancing detail with higher level concepts and approachability. It simultaneously and paradoxically makes ML feel like magic and demystifies the technology, particularly in the case with neural nets.

The pace is good throughout, there are forays into machine learning and AI history, discussions about the distinction between types of AI, computational progress and why it's spear-headed the AI revolution and finally what the future likely holds.

Highly recommend to any enthusiast or someone wanting to learn more about the technology.

[Abe Shocket · Rating 2 out of 5]

OK as an introduction, but you have to have some familiarity with data mgt, programming, etc. I don't know why it bothers me so much to see the word "nowadays", which usual shows up in the abstract of a technical paper and then turns me off immediately. This author got carried away with it and uses the word in practically every paragraph.

[Jessica Keener · Rating 4 out of 5]

Good high level introduction

[Dr. Tathagat Varma · Rating 2 out of 5]

A good coverage of the basic issues in ML

[Pat · Rating 3 out of 5]

Machine learning is a continuously evolving field, and any book that promises to be on its cutting edge will surely go out of date fairly quickly. This edition was published in 2021, and promised to be an introduction to machine learning for the non-technical. As a "concise overview of machine learning", it was fairly successful. It did an excellent job setting up the importance of data and the evolution of computing, covering supervised, unsupervised, and reinforced machine learning. By starting with simple linear regressions, it provided some theoretical underpinning for the volume of concepts to follow.

A couple of specific issues I took with the book centered around the chapters on neural networks and the future of machine learning. On neural networks, the author gave the impression that we understand what neural networks do. The whole premise of a neural network is that we don't understand what's going on; rather, the computer continuously updates and "learns" patterns in the data that we may not be able to detect, or even patterns that seem obvious but turn out to be anything but. An example of this is a recent experiment that found out a neural network, when left to its own devices, developed a complex form of adding two numbers that was completely unintuitive to humans. The author later talks about how networks are a "black box" model in Chapter 7, but by this point the damage is done. The confidence around how networks work in Chapter 5 seems a bit out of place.

In the chapter on machine learning's future, the author seemed way out of his depth. Most of the chapter is speculation, oscillating between describing how much data we will continue to produce in the future and one-off guesses about what technologies will be important in the future, and (relatively) how far off those technologies are. What struck me were not the technologies present (but some were doozies - flying cars?), but those that were conspicuously absent.

The whole book does a wonderful job referencing different papers for the curious reader to look into if they are interested. These are all very helpful, but one of the most important papers of the last decade is conspicuously missing: "Attention Is All You Need". This paper, defining the transformer architecture, has transformed the field of generative AI, especially in the area of large language models (LLMs). Not bringing up LLMs, or generative AI at all, is a bit of a failure of this book that is understandable but disappointing. GPT-3 had already been released by this point, as had DALL-E. Though they were not nearly as widespread as they are now, a paragraph mentioning them would have been helpful, as a reader with no prior knowledge picking up the book today would be a bit lost.

Overall, the book was quite bland for one with an "MIT Press" label. Even though it states in the introduction that it was a non-technical book, I was hoping for more of the mathematics behind much of the content. It is tough to gain much understanding of machine learning without understanding some of the fundamental math behind it; otherwise, you're mostly learning terminology, ideas, and other helpful facts that may help you in your job/business, but don't let you truly understand what you're working with.

I was also quite disappointed at the absolute lack of historical context for most of the book. Chapter 1 was a notable exception, providing some sense of the history of computation. But where were the specifics? Where was the story of Target predicting a teen girl's pregnancy before her father even knew? What about the early history of machine learning and AI, struggling with deterministic models before moving to others? The idea that neural networks were largely disregarded until Hinton et. al. brought them back into popularity? Any history (or even a separate focus) on statistics, the backbone of machine learning? Instead, the book presents concepts and ideas, with vague examples and future cases, all while preaching about the importance of data.

Overall, I think this is a good read if you don't have much interest in machine learning but need to be aware that certain concepts exist (like clustering or classification), or want to understand the very basics of a neural network. If you have any interest at all in machine learning, I'd recommend staying away and focusing on other books specific to the concepts you want to learn. Learn statistics from doing your own projects or other statistics-specific sources. Understand the fundamentals of reinforced learning by reading the AlphaGo paper. Learn the history of machine learning from other Silicon Valley or science/math related history books from the 20th century (I'd recommend "The Dream Machine"). Learn about deep learning using Bengio's or Nielsen's textbook. Heck, even give the author's own deep learning textbook a crack. But probably don't read this book.

[James · Rating 3 out of 5]

Machine Learning was a bit of a mixed bag for me. As others have stated this is a high-level conceptual approach to the subject. There is very little mathematical expression and it appears aimed at the layperson; however, the reader would be served by at least a fundamental understanding of probability and statistics.
The book is probably useful to management types or just the the random subject scanner who wants to know a bit more about the subject. Alpaydin introduces several problems and machine learning solutions. It's a very practical approach and I found it helpful in examining problems I face in my own work. If anything, it may provide you with a sense of different perspectives on systematically addressing problems.
I thought the writing was a bit disjointed. The front and back bookends read clearly to the layperson, while the middle makes some assumptions of prior knowledge. This bothered me only slightly because of how rudimentary the first section is. It started really basic, and at a certain point jumps to a different way of talking in domain specific parlance. It wasn't really a fault as much as it was not a smooth transition in my eyes.
In the end I found it interesting and it gets a pretty down the middle three-stars "I liked it".

[Marielena · Rating 3 out of 5]

This was a very informative introduction to Machine Learning and its applications in today's world. I found it indeed beginner-friendly, providing enough examples to clarify the topics discussed (I even felt that the introduction was a bit... too introductory). I was also happily surprised as I got to learn about a bunch of stuff that I encounter very often but wasn't aware of their origins (for example, about the Captcha test or the origins of the term "Data Mining"). Overall, I was rather pleased with it!

In the audiobook version, the part about different types of machine learning (supervised learning, reinforced learning, etc.) is a bit harder to get through as it is more technical. The narration doesn't help much in this respect - it's tolerable but rather monotonous. My brain tended to generate plenty of theta waves during listening (i.e., getting ready for nap time...)

Who recommended the book to me: no one - I was just looking for a decent book for an introduction to machine learning
I would recommend to: people interested in AI and in being introduced to the science of machine learning

[An Te · Rating 3 out of 5]

A helpful primer on machine learning. A near-complete introduction to the subject of machine learning from its applications, theories, tools, manifestations and its possible future. Some material is a rehash of his book "Data Science."

For a reader, one would be aware of the recommendations that come our way from websites/retailers, all generated from machine learning methods, as they may simply be reinforcing your preferences as we speak. The future methods will hope to add more diversity to our recommendations. It must be noted also, the book is pro-machine learning but many will still desire the privacy that a world governed by machine learning will certainly jeopardise or preclude entirely. The digital age has changed some of the infrastructure on how we connect, for good or ill. Machine learning would best be equipped by the devout and the tech-savvy. That much was lacking in this book. But on the whole, it covered all the technical ground.

[Peter Aronson · Rating 4 out of 5]

Three-and-a-half stars. A clear description of the various sorts of machine learning approaches, in as much useful detail as possible without resorting to the mathematics. That a really good thing.

The only drawbacks is that the author is somewhat careless of facts when discussing things outside of his topic (see the absurd statement about who got painted), which makes me wonder how careful they were when on topic, and that they're clearly a bit of a techno-utopian. Oh, they give the standard warnings, but his heart clearly isn't into them and he doesn't actually seem to be worried about how machine learning might be used in ways that make society worse.

So, a worthwhile book, but to be taken with a grain of salt about any non-technical aspects.

And, of course, since it was last revised in 2021, it is already out of date in some ways (like nothing about Large Language Models).

[Emory Scott · Rating 4 out of 5]

This book was a good conceptual overview on a wide variety of topics in machine learning. It was a great place to find points to explore in greater detail, but I would just as soon start by familiarizing myself with web articles. It covered a wide variety of topics, and i understand it’s difficult to condense such a large concept into a single book, but it seemed a bit all over the place at times - a key takeaways section at the end of every chapter would have been great. It helped me realize the purpose and point behind machine learning, but it is good to note this book is completely conceptual and does not dive into fránjalas detail of how the learning actually works or how to take actionable steps on your own. The use of graphs was beneficial in understanding topics, but I found it often had to GPT certain vocabulary terms to actually understand them. I’m excited to read the next installment on Neural Networks.

[Mark · Rating 3 out of 5]

I'm torn on my reaction to this. There will be a wide reaction to this based on the reader's expectations. If you are after learning about the algorithms or specifics of how machine learning works, you will likely be disappointed (which, admittedly, was my reaction because of my expectations and goals). If you just want an overview focused more on uses, history and where it may go, with only a little dipping into specifics, you will likely greatly appreciate this.

To me, it felt like a mixture of concepts, mostly at a high level, but not giving enough understanding to know why one algorithm is picked over others and in what contexts. Thus, I didn't get what I was personally wanting (possibly, through no fault of the author).

If your expectations are right, you'll like it, because the author clearly knows a lot, but it wasn't the "give me a methodical overview" that I was wanting.

[J. Joseph · Rating 5 out of 5]

Machine Learning is a quick and easily digestible primer for the what and why of machine learning, written in a way that anyone from a layperson to a professional can quickly learn. Importantly, it does not provide the how or any technical instructions on these systems, as that is not the aim of this informational book. It's divided into eight chapters: why we're interested in machine learning, statistics and data analytics, pattern recognition, neural networks and deep learning, learning clusters and recommenders, unsupervised learning, ethical and legal challenges and risks, and finally a look towards the future.

The chapters have such a natural progression for the concepts, with appropriate explanations that are clearly comprehensible but not condescending. Alpaydin was upfront about the intention that this is for laypersons first, professionals second, and this was excellently accomplished. As well, there are tons of useful and clear figures and images to explain the architecture of some systems or of the concepts, which is supplemented by a full glossary at the back of the book. From a critical perspective, all I want to raise is that some folks may feel this is too simplified at times, while too complex at others. I raise this only to flag it for others who are unsure what to expect, and not really as a true negative, since machine learning can itself fluctuate on its difficulty depending on topic.

[Antonis Maronikolakis · Rating 3 out of 5]

Machine Learning by Ethem Alpaydin is a short book on Machine Learning. It serves as an introduction to the field, explaining in a nutshell the different techniques and algorithms in Machine Learning. The author takes great care to bring forward the concepts in a simple manner so that newcomers to the field can get a taste of what to expect.

It does not go in depth on the specifics and it doesn’t introduce the mathematical background of the field, but it gives enough to entice and give a quick overview to the reader.

The book is not to be used as a textbook, but as a quick read on Machine Learning. Ideal for someone who wants to quickly get the gist of the field, or to someone needing an introduction.

[Simon Zuberek · Rating 5 out of 5]

Despite having been written with a lay audience in mind, the book delivers for a wide slice of readership. If you've been mesmerized (or frightened) by ChatGTP's antics and are curious to learn about the magic behind the technology, this position will establish a solid foundation from which one can begin exploring the topic. If you are about to take machine learning at a university, this short text will serve as a solid primer for the course. If you've already taken machine learning, it will help you systematize your knowledge into a coherent, holistic whole. And if you're already a proficient ML scientist, the text will teach you how to explain some of what you do to your family and friends. Overall, and excellent read.

[Andrew Miller · Rating 5 out of 5]

If you're interested in learning more about machine learning, then this book is definitely worth checking out. It provides a comprehensive overview of the field and is a great starting point for anyone looking to dive deeper into this exciting area of study. But to truly succeed in machine learning, it's crucial to have high-quality annotated data sets. This article https://www.digitalconnectmag.com/how... offers valuable insights and guidance on selecting the best tools for your data annotation needs. By using the insights and tools provided there, you'll be able to optimize the performance of your machine learning models, and fully realize the potential of the knowledge presented in the "Machine Learning" book.

[CJ Spear · Rating 2 out of 5]

Who is this book for? Fools like me, I guess. I do not believe a general reader looking to gain a layman’s understanding of ML would benefit from this book. It is simply too much information too poorly communicated. Conversely, I also do not believe a student looking to learn ML would benefit from this either. Without examples and coding exercises and more fleshed out chapters, this information is not useful and a student will likely have to relearn all of the concepts brushed over in this book.

The person writing this seemingly did so without any passion for communicating the ideas behind machine learning. It felt like a droning lecture. His lack of inspiration made this book difficult to finish.