Learning The Math Behind Deep Learning

Ever wondered what magic Deep learning does behind the scene. How the math behind Deep learning works. We still don't know exactly how these algorithms work but knowing the math behind it definitely helps and makes us understand it one step further.
For learning the math behind Deep learning you first have to decide what you want out of Deep learning. A number of scenarios could be :

1. A someone who just wants to do integrate Deep learning into his application and is curious about what it is - then you don't need to get into the maths part.
2. Someone who wants to do Deep learning casually. Don't want to get into specific details of the deep learning - then getting the basic intution will work for you.
3. Someone who wants to go Deep learning seriously and then maybe do research in deep learning - then you the more you do is better.

So, take a moment and decide what you want to do after learning Deep learning and then proceed further to the following sections and strategize what's best for you

Calculus

• The Essence of Calculus series link

This series introduces you the intuition behind the calculus. Watching this series will be most useful if you have studied calculus and don't know the intuition behind it but if you haven't studied calculus, I would recommend you to buy a book on calculus and solve some questions on it.

• MIT Multivariable Calculus link

If you want to go deeper and learn about the multivariable calculus. It's very intuitive if you already know calculus.

Linear Algebra

• The Essence of Linear Algebra series link

You must take this course for deep learning. This series will give you almost all the fundamental intuition you need for the linear algebra part but this series doesn't involve solving questions on linear algebra. So if you new to linear algebra then I would recommend you try solving some questions. There are some good online resources where you can find questions to solve. Just google it.

• MIT Linear Algebra link

If you are serious about deep learning, I strongly recommend you take this course. I would also recommend you to buy the book given on the resources tab on this page and solve the questions given in the book. Linear algebra is the most important subject for deep learning.

Probability and Stats

• Khan Academy Probability and Stats series link

This is really good series. They are short and to the point and on khan academy site you also get to solve some practice questions. So this is a must take for deep learning.

• Harvard Stats 110 link

Blitzstein does a really good job explaining the intuition underlying the main concepts. I would recommend this over the MIT's stats course, as this course tackles the concepts with a more intuition.

Bonus Material

Convex Optimization link