brain of mat kelcey...
my updated list of cool machine learning books
November 01, 2020 at 09:40 PM | categories: Uncategorizedawhile ago i posted my list of cool machine learning books, but it's been awhile so it's probably time to update it...
Mathematics for Machine Learning by Marc Peter Deisenroth, A. Aldo Faisal & Cheng Soon Ong.
this is my personal favorite book on the general math required for machine learning, the way things are described really resonate with me. available as a free pdf but i got a paper copy to support the authors after reading the first half.
Linear Algebra and Learning from Data by Gilbert Strang.
this is gilbert's most recent work. it's really great, he's such a good teacher, and his freely available lectures are even better. it's a shorter text than his other classic intro below with more of a focus on how things are connected to modern machine learning techniques.
Introduction to Linear Algebra by Gilbert Strang.
this was my favorite linear algebra book for a long time before his 'learning from data' came out. this is a larger book with a more comprehensive view of linear algebra.
Think Stats: Probability and Statistics for Programmers by Allen Downey.
this book focuses on practical computation methods for probability and statistics. i got a lot out of working through this one. it's all in python and available for free. ( exciting update! as part of writing this post i've discovered there's a new edition to read!)
Doing Bayesian Data Analysis by John Kruscgke
on the bayesian side of things this is the book i've most enjoyed working through. i've only got the first edition which was R and BUGS but i see the second edition is R, JAGS and Stan. it'd be fun i'm sure to work through it doing everything in numpyro. i might do that in all my free time. haha. "free time" hahaha. sob.
The Elements of Statistical Learning by Hastie, Tibshirani and Friedman
this is still one of the most amazing fundamental machine learning books i've ever had. in fact i've purchased this book twice and given it away both times :/ i might buy another copy some time soon, even though it's been freely available to download for ages. an amazing piece of work.
Probabilistic Graphical Models by Daphne Koller & Nir Friedman
this is an epic textbook that i'd love to understand better. i've read a couple of sections in detail but not the entire tome yet.
[Pattern Recognition and Machine Learning] (https://www.microsoft.com/en-us/research/people/cmbishop/prml-book/) by Christopher Bishop
this is probably the best overall machine learning text book i've ever read. such a beautiful book and the pdf is FREE FOR DOWNLOAD!!!
Machine Learning: A Probabilistic Perspective by Kevin Murphy
this is my second favorite general theory text on machine learning. i got kevin to sign my copy when he was passing my desk once but someone borrowed it and never gave it back :( so if you see a copy with my name on the spine let me know!
Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow by Aurélien Géron
this is the book i point most people to when they are interested in getting up to speed with modern applied machine learning without too much concern for the theory. it's very up to date (as much as a book can be) with the latest libraries and, most importantly, provides a good overview of not just neural stuff but fundamental scikit-learn as well.
Machine Learning Engineering by Andriy Burkov
a great book focussing on the operations side of running a machine learning system. i'm a bit under half way through the free online version and very likely to buy a physical copy to finish it and support the author. great stuff and, in many ways, a more impactful book than any of the theory books here.
Introduction to Data Mining by Pang-Ning Tan, Michael Steinbach & Vipin Kumar
this is another one that was also on my list from ten years ago and though it's section on neural networks is a bit of chuckle these days there is still a bunch of really great fundamental stuff in this book. very practical and easy to digest. i also see there's a second edition now. i reckon this would compliment the "hands on" book above very well.
Speech and Language Processing by Dan Jurafsky & James Martin
still the best overview of NLP there is (IMHO). can't wait to read the 3rd edition which apparently will cover more modern stuff (e.g. transformers) but until then, for the love of god though, please don't be one of those "this entire book is irrelevant now! just fine tune BERT" people :/
[Numerical Optimization] (https://link.springer.com/book/10.1007/978-0-387-40065-5) by Jorge NocedalStephen J. Wright
this book is super hard core and maybe more an operations research book than machine learning. though i've not read it cover to cover the couple of bits i've worked through really taught me a lot. i'd love to understand the stuff in this text better; it's so so fundamental to machine learning (and more)
Deep Learning by Ian Goodfellow
writing a book specifically on deep learning is very dangerous since things move so fast but if anyone can do it, ian can... i think ian's approach to explaining neural networks from the ground up is one of my favorites. i got the first edition hardback but it's free to download from the website.
Probabilistic Robotics by Sebastian Thrun, Wolfram Burgard and Dieter Fox
when i first joined a robotics group i bought a stack of ML/robotics books and this was by far the best. it's good intro stuff, and maybe already dated in places given it's age (the 2006 edition i have) but i still got a bunch from it.
TinyML by Pete Warden & Daniel Situnayake
this was a super super fun book to tech review! neural networks on microcontrollers?!? yes please!
Evolutionary Computation by David Fogel
this is still by favorite book on evolutionary algorithms; i've had this for a loooong time now. i still feel like evolutionary approaches are due for a big big comeback any time soon....
in the mail...
the good thing about writing a list is you get people telling you cool ones you've missed :)
the top three i've chosen (that are in the mail) are...
Causal Inference in Statistics by Judea Pearl, Madelyn Glymour & Nicholas P. Jewell
recommended by animesh garg who quite rightly points out the lack of causality in machine learning books in the books above.
Information Theory, Inference and Learning Algorithms by David MacKay
i've seen this book mentioned a number of times and was most recently recommended by my colleague dane so it's time to get it.