I don't have to jargon around how AI has taken over the world.The world is changing at a rapid pace and you gotta keep up with it or become a fucking dinosaur and disappear. It is time that AI education becomes more formalized and accessible to wide range of people and trust anyone with a sane(this is subjective) mind can start with Machine Learning and have a career in it.
Here is my guide for you to begin with Mathematics for Machine learning and explore its possibilities(still in research tho)
The Math Tragedy - Now, when it comes to Machine learning the first barrier to entry is mathematics and people universally suck at it .It is by far the most feared reason of why people don't get into Machine learning in the first place and i wouldn't deny that its tough and really integral to learn when you want to understand anything in ML. Now a lot of advises around Quora tell you that you don't have to learn a lot of math to get started with ML,which is true but you cannot sustain in the ecosystem where Research papers are coming out so fast and keeping up with it would require a lot of mathematics.
In high school i pretty much sucked at Math because I could not relate to it,there
was no intuition and purpose of maximizing a function over a domain or finding
inverse of a matrix.They were mere problems from text book that i had to do so
that i could crunch some marks(still a big deal in today's education).But over the
last few years i have started enjoying math and understanding how anything that ever
existed has mathematical feel to it and why you would maximize a function in a
given situation. My main point is , now you have the CONTEXT for learning it
and you will be more comfortable and moreover appreciate it.
So lets get started with the listicles (tried to not make it sound like testicles) . Most of the math you would need are-
1.Linear Algebra : Its every where in ML and Deep learning , kinda like the crux of a lot of algorithms and operations very important to get hang of it
MIT 18.065 Matrix Methods in Data Analysis, Signal Processing, and Machine Learning, Spring 201 , Instructor: Gilbert Strang - This is a fresh course from the legend Gilbert Strang,where he actually condenses down the entire linear algebra just for data science and its all that you need.
Check it out here -
3blueOneBrown Essence of linear algebra - One of the best ways you can learn and most importantly understand math.
Check it out here -
- The Deep learnin book by Ian goodfellow Ian Goodfellow and Yoshua Bengio and Aaron Courvill (The pioneers in AI) - This book has a prerequisite section in beginning which covers pretty much everything you need to study in mathematics to get a knack. Chapter 2 of this book has pretty much every thing.BTW this book is the bible for deep learning and most importantly its free !
Check it out here -
Khan academy - This guy is a genius and yet he is named Salman Khan(i am not even making this up ! ).He has a very organized way of teaching which is very easy to understand for those who like hand holded way of teaching.
Check it out here -
This is enough content for you to nail Linear Algebra.
Try to imagine things while you learn it as that will make it easier for you to remember and most importantly understand it.
Now lets come to the biggy -
Probability and Statistics **- Probability and Statistics are huge topics where folks in fact pursue there entire graduate as well as undergraduate studies,but you don't have to go that deep.Understand Probability and Statistics in correlation(i couldn't help it) with your Data.Data is the center of everything that you will ever do and you better fucking understand it well.
Harvard's Statistics 110 - This guy is a brilliant , well versed statistician and very good communicator.He is very rigorous with the math path but also explains the concepts really well.He covers all the complex topics you need to cover.
Check it out here -
- Professor leonard's Statistics - I discovered him and his abs few months ago and his channel has all of the mathematics that you will ever need.He explains things really well and at a pace that anyone as dumb as they get can understand.
- Statistics - The Art and Science of Learning from Data 3e - Agresti, Franklin - I found this book a week before and i just wished i had found this one before.This book is brilliant in ways that it gives you all the practical scenarios where you would use a statistical concept and not just mathematical derivation.This is a must if you want to do well when understand your data (Exploratory data analysis).
Download it for free -
- Khan academy Statistics - No list can end without these.
Now coming to something that you will scratch your head with the most is OPTMIZATION THEORY and its basically calculus from your high school but with context.
Optimization is crux of all the learning algorithms and you need to take it seriously.
- 3Blue1Brown's Calculus - After watching this series you would think that you could have invented calculus by yourself all alone !
- Professor leonard's Calculus 3 - A brilliant tutorial to get started with 3D calculus
- Imperial College of london's Mathematics for Machine Learning: Multivariate Calculus - The name is enough
I will call it a wrap.All these resources are all that you need as a prerequisite to machine learning.
I will come up with more articles,resources and in the future,with some interesting tutorials. So stick with me.