If you’re about to start your first Machine Learning course and wanted to know how much maths you’re gonna need, then good news! I’ve compiled a list of resources which I think will be useful in getting you up to speed, assuming you’ve studied lower undergraduate maths.
The list, inspired by Prof. Iain Murray’s own list, is split by each major subfield of mathematics that is required in machine learning, i.e. probability theory, linear algebra, and calculus.
General
Before you start, you might want to assess how badly you’ve forgotten all the maths you’ve put effort into learning not so long ago. So, first read through this cheat-sheet put together by Prof. Iain Murray:
By skimming through it, you’ll identify concepts and formulas you have forgotten or are not comfortable with. These will constitute your weak points. Now let’s strengthen those soft spots.
Probability
- Basic probability tutorial. This tutorial is long and awesome. Do go through it all. Go through all the examples, and solve all the exercises. It will give you a better intuition about important concepts like Bayes’ Rule or marginalization, for instance. Special thanks to Prof. Sharon Goldwater for redistribution permissions.
- Mathematics for Machine Learning, Chapter 6: Probability and Distributions. More detailed and advanced than the tutorial above. You will most likely encounter some of the later bits in your Machine Learning course, so I would not bother too much. It is far more important to develop an intuition about the basics.
MIT Introduction to Probabilities MOOC. Part I, the fundamentals. If you prefer video.
Linear Algebra
3blue1brown’s Essence of Linear Algebra video series. Entertaining videos; relatively short and sweet.
Mathematics for Machine Learning, Chapter 2: Linear Algebra.
Mathematics for Machine Learning, Chapter 3: Analytic Geometry. Not as essential as the other two, but it can still offer some intuition on important topics like distances and translations.
Calculus
- 3blue1brown’s Essence of Calculus video series.
- Mathematics for Machine Learning, Chapter 5: Vector Calculus.
Conclusion
There is quite a lot of material to cover, so make sure to take plenty of breaks so that you do not lose focus and get bored. Do a bit every day, instead of cramming everything in a weekend. Good luck!