You will have guessed by now that I am keen to write more frequently about an area of expertise which is close to my heart: digital transformation and how to demystify it.
About the author
Pedro Domingos researches machine learning and data-mining with a mission to
“Make computers do more with less help from us, learn from experience, adapt effortlessly, and discover new knowledge”.
He is also professor of computer science and engineering at the University of Washington.
About the book
This book contains the most painstaking research into machine learning that I have ever read. Domingos wastes no time in defining machine learning precisely, a welcome change from the noise on the subject that we hear around us these days.
For him, machine learning is a set of algorithms that figure things out on their own, by making inferences from data. The more data the algorithms gather and infer, the better they get. They are not programmed, but programme themselves instead.
Until now, machine learning has been considered as a new technology that builds itself.
For long, people repeated the famous Picasso sentence “computers are useless, they can only give you answers” referring to the lack of creativity in computers until came machine learning
This book answers critical questions currently being debated such as:
Domingos splits his book into ten chapters that address these questions through a framework consisting of five pillars of machine learning: inductive reasoning, connectionism, evolutionary computation, analogical modelling, and Bayes’ theorem.
The book also offers a view of the future where machine learning algorithms comprehend how the world operates and how humans behave. Since his book was published, there has been much progress, with intervening academic journals suggesting that his vision may not be that unrealistic.
Yet the book does have some limitations. Machine learning may be misused as a technology that is not fully understood. This is why the book urges everyone to attempt to understand machine learning. For example, politicians may use knowledge derived from machine learning to make unreliable promises to voters.
For me, the greatest value of the book – contrary to some critiques about its lack of technicality – is the focus on the business use of machine learning. The book’s differentiator lies in the translation of technical concepts into accessible, logical, business-relevant processes.
Here Domingos suggests various ways of making machine learning more fit for use
This book positions data as a strategic asset. It shows how machine learning allows individuals, businesses, and governments to operate more successfully. It equips consumers to know what data is available that others do not have (and vice versa) and how to make the best use of available data.
In 2016, Bill Gates recommended the book as one of two that everyone should read to understand AI. This book’s mission, put simply, is to acquire past, present, and future knowledge from data by a single and universal learning algorithm.
For those who prefer to learn through videos, here is an insightful talk for the author about his book.
Who should read this book
This book is relevant for readers focused on the business use of machine learning, as mentioned earlier. Yet, it is also a highly recommended read for
And everyone who wants a primer on machine learning!
This post also appeared on LinkedIn.
This is a blog post about managing one's career effectively. A post made possible by our favorite contributor Ken McKellar!
This book contains the most painstaking research into machine learning that I have ever read.
Organizations do no longer have the luxury to refrain from going digital if they want to “survive and thrive”; it is a one-way ticket towards the creation of economic value, agility, and speed. The book is a transition for digital transformation: from widely discussed to widely understood.
In turmoil and when the rubber meets the road, it is time to remember the ground rules for managing money: save some, spend some, and give some.