Machine learning remains a hot topic as organizations attempt to squeeze insights and competitive advantage from their data sets.
I’ve selected five books that may be of interest if you are embarking on a machine learning initiative or career. Obviously, there are much more great books out there and I’d like to hear your recommendations—recommendation engines do such a terrible job…
Note that machine learning is a fast moving discipline, so tools and techniques are being revised and replaced constantly. Scan the reviews for books for indications that they may have passed their sell-by date. This is particularly a problem with books that focus on a given technology.
Books and articles on machine learning and big data tend to fall into one of two camps. There are the hype-fueled business articles on one side and the technical mathematics and computer resources on the other side. There’s little available in the middle ground.
Pedro Domingos’ “The Master Algorithm” is a valiant attempt to bridge the chasm. As a machine learning practitioner, I welcome a book that I can recommend to potential clients. Informed clients have a better idea of the potential and limitations of machine learning—and come prepared with ideas about where the technology could be used in their organizations.
Five machine learning “feeder” disciplines are covered in the book:
R is the darling of the data science community. It’s a functional language and environment for statistical computing and graphics. It’s open source, fast growing and has recently attracted the attention (and resources) of Microsoft. There are many machine learning packages available for R.
Cory Lesmeister’s “Mastering Machine Learning with R” gives a solid introduction to applying the most popular machine learning techniques using R. If you want to get “hands on” with machine learning, R, and this book, would be a good place to start.
You’ll need some knowledge of R to get the most from the book. Jared Lander’s “R for Everyone” is a good introduction to the language and environment.
At some point, possibly very early, in your machine learning career, you’ll be confronted with big data. At the time of writing, Apache Spark is the leading big data machine learning solution. Tools like Azure ML are more immediately accessible, but Spark is the poster child.
Nick Pentreath’s “Machine Learning with Spark” is a good place to start learning about Spark’s machine learning libraries. Note, however, that Spark is a very fast moving project, and a lot has changed with the recent launch of Spark 2.0. This book covers the “old” libraries, but you can build on the knowledge gained when moving to the later versions of Spark.
Publishers Packt have a revision of this book due in early 2017.
If there’s a bible of machine learning, Kevin Murphy’s textbook “Machine Learning: A Probabilistic Perspective” is probably it. It’s a standard text in many top university courses.
The book is extensive, weighing in at almost 1100 pages, and is a textbook. It’s detailed and not light reading. But, if you want to gain a deep understand of machine learning techniques—the mathematics behind them—this is the place to go.
Much has happened in the machine learning space since the book was published in 2012, so it’s not up to date with some of the latest techniques and thinking—but the core material still applies. Apparently, Murphy is working on a major revision.
As you can’t fail to have noticed, the apocalypse is no longer nuclear Armageddon—it’s being enslaved or eliminated by evil killer robots. We’re not sure if our future is being hunted down by Skynet or being inserted into the Matrix—but we’ve been assured it’s grim.
Technology and scientific luminaries such as Bill Gates, Stephen Hawking and Elon Musk believe AI is the greatest existential threat facing mankind. Once machines can out-think us it’s game over.
Nick Bostrom, the author of “Superintelligence”, is a philosopher and director of the Future of Humanity Institute at Oxford University. His work is credited with bringing many of these concerns about AI to the fore.
Anyone involved in machine intelligence work will occasionally encounter resistance from people who have been exposed to these ideas but are not familiar with the more mundane day-to-day uses of the current state-of-the-art in machine learning. It pays to be informed about the arguments on both sides and knows how to position practical machine learning in the content of concerns about superintelligence.
If you are interested in machine learning and want some hands-on training, you may wish to consider the following Learning Tree courses: