# Math For Deep Learning — Do I Need It?

“Math is the language of the universe. So the more equations you know, the more you can converse with the cosmos.” — Neil deGrasse Tyson

So you want to start learning Deep Learning or you’re already learning it. Or, maybe you are not sure if you should learn it because you think you’re not the best when it comes to Math. So, you’re scrounging the Internet to figure out if you need Math in order to learn deep learning or machine learning. Well, I am here to tell you that you absolutely don’t need Math to get into deep learning. Also, the previous sentence was a lie.

You do need some Math. Question is, what parts of Math do you need?

Now, before you get all tensed up about all this because some guy on the Internet just told you that you need Math, relax. You do need Math, but you don’t need all of it. Also, I’m going to tell you the secret sauce at the end of this article. But first, what kind of Math do you really need?

1. Linear Algebra: This is one part of Math that you absolutely need to be familiar with. Even if you only want to be a deep learning practitioner, and not a researcher, you still need linear algebra. Why? Because almost all your data will be in the form of multi-dimensional matrices. And a lot of magic that you will be doing using your code will require you to understand the operations on such matrices. This is absolutely something you should learn.
2. Differential Calculus: This one is not absolutely necessary if you just want to create some fun projects in deep learning. But, if you want to go deeper and understand how things work or you want to get into some research, then you will need this tool in your toolbelt.
3. Statistics: Since you will be dealing with a lot of data and you will also need to manipulate, understand and visualize that data, you will need statistics.
4. Probability: If you go into some serious applications, then you will also need probability because with any kind of deep learning, you will be dealing with probabilities. Serious applications such as self-driving cars, artificial intelligence etc. will require you to work with probabilistic models. But for some lightweight deep learning, you won’t need much of this.

This is pretty much all you need for deep learning, in terms of Math. If you just want to play around with deep learning and do some lightweight projects for fun, then you will be fine by having just some familiarity with those concepts. But, if you want to go into full research mode, then you will need to be quite well-versed with these things.

Also, you don’t need to be Math wizards to be deep learning practitioners. You just need to learn linear algebra and statistics, and familiarize yourself with some differential calculus and probability.

And now, as I promised, the secret sauce. If you feel intimidated by Math and that is your only reason for not getting into deep learning, or any other field, then this secret sauce is for you. And the secret sauce is this:

The only way to learn Mathematics is to do Mathematics — Paul Halmos

Where can you learn Math? You can use Khan Academy, Udacity, and Fast.ai(for computational linear algebra). You can also use the Deep Learning Book and go through its unit-1 to learn all prerequisites for deep learning.

Here’s a small video for you by Khan Academy to motivate you to learn.

I hope you don’t get intimidated by Math or by learning, and really get into the field of deep learning. And I will leave you with one of my favorites quotes.

Dare I say, we can change the world — Bill Nye