No. 117 of 500 No. 12 in Machine Learning

Mathematics for Machine Learning

By Marc Peter Deisenroth, A. Aldo Faisal and Cheng Soon Ong

Published
April 23, 2020
Print length
398 pages
Reading level
Intermediate
ISBN-13
978-1108455145
Publisher
Cambridge University Press
Written for
Students and practitioners preparing for graduate machine learning courses or rigorous ML textbooks

Synopsis

Marc Peter Deisenroth, A. Aldo Faisal and Cheng Soon Ong wrote this textbook for readers who discover that machine learning books assume mathematics they never fully learned, or learned in scattered courses with clashing notation. Part I gathers the essentials in one consistent framework: linear algebra, analytic geometry, matrix decompositions such as eigendecomposition and the singular value decomposition, vector calculus including backpropagation and automatic differentiation, probability and distributions, and continuous optimization. Each chapter introduces concepts with minimal prerequisites and supplies worked examples and exercises.

Part II puts those tools to work by deriving four central methods from first principles: linear regression, treated through maximum likelihood and Bayesian inference; dimensionality reduction with principal component analysis; density estimation with Gaussian mixture models and the expectation-maximization algorithm; and classification with support vector machines. A bridging chapter on how models meet data frames training, evaluation and model selection. The authors keep an updated PDF free on the book's website, along with Jupyter notebook tutorials. Readers come away able to follow the derivations in advanced texts, although the book deliberately stops short of deep learning architectures.

Scorecard

Editorial score

How it scored

Editorial score: 82.1 / 1005 weighted criteria

  • Influence30%

    25.5/30
  • Authority25%

    21.3/25
  • Practical value20%

    16.0/20
  • Currency15%

    12.8/15
  • Readability10%

    6.5/10
  • Editorial score

    82.1/100

How the rank was computed

Rank No. 117 of 500 comes from a composite of 61.7: observable signals 24.8 of 55 points plus editorial review 36.9 of 45 points.

All 11 inputs
InputPoints
Observable signals24.8/55
Reach (Amazon ratings volume)11.1/16.5
Consensus across best-of lists5.5/13.8
Awards and bestseller evidence0.0/8.3
Reader rating, sample-adjusted6.8/8.3
Amazon sales rank1.4/5.5
Recency0.0/2.8
Editorial review36.9/45
Author authority9.6/11.3
Influence11.5/13.5
Practical value7.2/9.0
Readability2.9/4.5
Currency5.7/6.8

Each criterion is scored from 1 to 10 and weighted, and the points add up to the editorial score. The rank combines observable signals (55 percent) with this score (45 percent). Read the methodology.

Verdict

A rigorous, free bridge from undergraduate mathematics to machine learning, anchored by four fully derived core methods.

Strengths

  • One consistent notation across all core topics
  • Free PDF plus Jupyter notebook tutorials
  • Applies the mathematics to four real methods

Keep in mind

The exposition can be terse, introducing results as needed, so readers without prior exposure may need a gentler companion text.

Key takeaways

  1. Vectors, matrices and their decompositions are the shared language of machine learning, so fluency in eigendecomposition and SVD pays off everywhere.
  2. Gradients of vector-valued functions, computed efficiently through backpropagation and automatic differentiation, drive the training of nearly every modern model.
  3. Many learning problems reduce to optimization, and understanding gradient descent, constraints and convexity explains why some models train reliably.
  4. Deriving linear regression, PCA, Gaussian mixtures and support vector machines by hand shows how probability, geometry and optimization combine in practice.

Notable facts

About the author

Marc Peter Deisenroth, A. Aldo Faisal and Cheng Soon Ong

Marc Peter Deisenroth is the inaugural Google DeepMind Chair of Machine Learning and Artificial Intelligence at University College London and previously taught at Imperial College London. A. Aldo Faisal leads the Brain and Behaviour Lab at Imperial College London, with appointments in bioengineering and computing. Cheng Soon Ong is a principal research scientist at CSIRO's Data61 in Canberra and an adjunct associate professor at the Australian National University.

Questions

What does Mathematics for Machine Learning cover?
Mathematics for Machine Learning has two parts. Part I covers linear algebra, analytic geometry, matrix decompositions, vector calculus, probability and distributions, and continuous optimization. Part II applies that material to derive four core methods: linear regression, principal component analysis, Gaussian mixture models and support vector machines. Every chapter includes worked examples and exercises, and the authors provide Jupyter notebook tutorials on the companion website.
Is Mathematics for Machine Learning free?
Yes. The authors keep a PDF of Mathematics for Machine Learning freely available for personal use at mml-book.github.io, including a version matching the printed Cambridge University Press edition and an updated version. Redistribution and resale are not permitted. Cambridge University Press sells the print edition, and instructors can request a solutions manual from the publisher.
What background do I need for Mathematics for Machine Learning?
The book assumes mathematics commonly taught in high school, such as derivatives and basic vectors, and introduces most concepts with minimal prerequisites. In practice, readers with some prior exposure to linear algebra or probability progress fastest, because the presentation is compact and brings in results as needed. Those starting from scratch can succeed by working the exercises and pairing chapters with video lectures or a gentler introduction.

Sources

As an Amazon Associate, Best AI Books earns from qualifying purchases. Best AI Books is sponsored by Iternal.ai, whose founder and CEO, John Byron Hanby IV, wrote The AI Strategy Blueprint, our #1 ranked book. How we rank books.