The world’s leading publication for data science, AI, and ML professionals.

The Math you need for Machine Learning

A list of resources that will help you level up with the math required for machine learning.

6 resources that will help you level up.

Photo by Roman Mager on Unsplash
Photo by Roman Mager on Unsplash

I never cared much about Machine Learning.

If we were playing the blame game, I’d certainly point to the "math is not my thing" excuse. I had seen it with my own eyes, and it seemed daunting.

Back then, we had to write training loops from scratch, beg large universities for cluster time, and deal with parallel libraries and remote debugging.

That was a long time ago.

Fast forward a few years, and I came around and gave it a try. To my surprise, I was more than ready to get into it!

The field had changed. The Math I needed was far from the scary wolf that everyone made me believe it was, and I’ve never had an issue with it.

A quick anecdote

It was 2015 right when I decided to pursue a Master’s.

Most of us went in hearing horror stories about the complexity of derivatives and how ugly linear algebra was. To our surprise, most who dropped out sucked as programmers, not as mathematicians.

That’s right: the lack of programming skills was a much bigger hurdle than anyone anticipated. The math, not so much.

Too much from one side

There are many papers, books, videos, tutorials, and all sorts of universities teaching about machine learning.

Do you know what all of these have in common?

Most of them are taught by researchers that have spent their lives looking at the field from the same angle.

This is not necessarily a bad thing, but it’s definitely one-sided.

When was the last time that you wanted to learn something, and you didn’t have to wrestle with the long-ass formula in the second paragraph of the article?

Math is important, but it’s not the only way to communicate. There’s a time when all we need is a metaphor that drives home a point. Sometimes analogies do wonders and change minds. Well-written essays explaining a concept using layman’s words are effective tools. And math…well, sometimes math is the way. It’s just not the only way.

A different approach

The funny part is that the math required by many machine learning programs can be taught as part of the course itself!

Nobody wants to start their journey with a detour. Where do you even start? More importantly, where does it end? How much of it is enough?

Learning what you need right when you need it is a different approach. We know how to operate this way. We have been improving ourselves for centuries taking things one step at a time.

Take that first step and cross the next bridge when appropriate.

Links that helped me

As soon as you decide to dive in, you’ll need some recommendations. The math for machine learning mainly centers around three topics:

  • Probability and Statistics
  • Linear Algebra
  • Multivariate Calculus

Although most of the content I consume comes straight from Google and YouTube searches, a more structured approach is incredibly helpful. That’s where these recommendations come in.

I picked two resources for each topic. The first should be easier to follow, and the second should offer a more in-depth perspective on the subject. However, all of them are incredibly approachable, and you shouldn’t have any problems following along.

Here is the list:

  • Seeing TheoryAn interactive website that takes you through some of the most critical probabilities and statistics concepts. These should be enough to get you started, and you will have fun while going through it!
  • Statistics 110: Probability – If you are looking to get a more advanced overview of Probabilities and Statistics, this course from Harvard University is an excellent introduction to probability as a language and a set of tools for understanding statistics, science, risk, and randomness.
  • Essence of Linear Algebra – Who doesn’t love Grant Sanderson’s YouTube videos? Go through this playlist for a refresher in Linear Algebra, and you’ll be more than ready to face any machine learning demons.
  • Linear Algebra – This is MIT Course 18.06, taught by Profesor Gilbert Strang. Definitely one of the best linear algebra courses that you’ll ever find. Prof. Gilbert makes the subject ridiculously simple and engaging.
  • Essence of Calculus – This is Grant Sanderson’s excellent take on calculus. A series of videos that are informative and make calculus feel like something you could have discovered yourself.
  • Multivariate Calculus – This is a free, beginner-friendly introductory course to building your confidence and introduce you to the multivariate calculus required to build many common machine learning techniques.

A final word

Give machine learning a try.

Many people believe that you either understand how electrons move in a wire or have no business changing an outlet. And that’s fine. You don’t need them to believe otherwise.

Don’t worry about the things you think you’ll need. The time for more will come, and you can think about them at that point. Start with something simple, and find your way through more of it as you feel you need it.

You’ll be fine.


Related Articles