Mathematics of Machine Learning

Do you need to know math to do machine learning? Yes! The big 4 math disciplines that make up machine learning are linear algebra, probability theory, calculus, and statistics. I'm going to cover how each are used by going through a linear regression problem that predicts the price of an apartment in NYC based on its price per square foot. Then we'll switch over to a logistic regression model to change it up a bit. This will be a hands-on way to see how each of these disciplines are used in the field.

Code for this video (with coding challenge):

Please Subscribe! And like. And comment. That's what keeps me going.

Want more education? Connect with me here:

Sign up for the next course at The School of AI:

More learning resources:

Join us in the Wizards Slack channel:

And please support me on Patreon:
Signup for my newsletter for exciting updates in the field of AI:

[kalebbruwer]

All those people who say, "Math is useless and you don't use it in the real world" are about to be replaced by machines running on all the types of math people tend to hate.

[hammadshaikhha]

Yo Siraj, this was a awesome and simple video. I think you have figured out the best combination of pacing, math, animations, and memes now. I would appreciate if you made more videos like these on topics like decision trees, SVM, back propagation, etc. I know you have covered these topics before but I feel the quality of videos now is a lot better for learning. Like there is a huge difference between this video which you can really learn from and your "backpropogation in 5 minutes".

Anyways I will submit for the challenge for this week, looking forward to other submissions as well.

[klopkerna]

You didn't talk about complicated Hilbert space. I'm so disappointed. 😥

[tanbirsohail]

This is my favourite channel to get introduced to everything in machine learning.

[tylerangert]

your video editing is getting much better :) i also can really tell you are working on your pacing and presentation tone. very good content.

[umakanetkar]

Brilliant video! Have been studying Machine Learning for a while but never knew what foundations it has been built upon. Thank you!

[darrellwoodwood]

Thank you Siraj for the clear, accurate, concise and enlightening overview of machine learning. It answers my most fundamental question that is always on my mind, viz., Why am I doing what I am doing? Thanks again!

[tejasvagarwal]

Please include this video in your playlist of maths of intelligence. I love that playlist and watch it once in a month.
Cheers

[nickjoseph]

"Like a nice... bowl." Killed it. Great job, Siraj!

[ozland]

I am literally enlightened on this subject.. i studied maths but never made sense so much! thank you!!!

[michaelfalkenberg]

Am currently learning Python to assist in corpus analysis within linguistics, everytime i get bored or demotivated I watch one of your videos to see what it is that programming can do and I get back to work. Thanx.

[camilocarrillo]

this guy is not crazy at all, after 1 day of watching videos(far more fun and efficient than articles) I understand almost everything, say 80% being ML calculus and back propagation the hardest. That being said I know nothing about calc so it's a bit harder for me to grasp the concept, anyone familiar should have no problems. Next challenge is to fully understand the calcs behind and then python the stuff. Great channel!

[vishalmishra]

How about the mathematics of all the new stuff - deep-learning/neural-nets, CNNs, RNN/LSTM, transformers/attention-networks, GPT-3, Alpha-Fold2 etc. ?

[andrewzelin]

Hi - a very informative and succint video of how these 4 areas of maths come together to lead to Machine Learning - thanks, Siraj! I am now able to use back-proagation to find a solution to a predicive / regession model! Happy to share my code!

[SabbirAhmedSibli]

Summery of the ML Stanford course by Mr. Ng :) Nice explanation btw..

[mind.journey]

Subscribed! My problem is that each of these fields is really vast, in fact you could spend years and still not be done learning statistics or linear algebra.
Could someone point out some good resources to learn each one of these disciplines?

[valoredramack]

So this Fall Quarter 2019, I'm getting ready to finish Calculus I in college which taught me how to work with Toolkit Functions, Derivatives, Limits, Domains, Ranges, Power Rule, Quotient Rule, Product Rule, Chain Rule, L'Hopital's Rule, and Logarithmic Properties. My question is....do I need to take Calculus II and III to be able to understand the algorithms used in Machine-Learning? Because I would much rather be taking more programming courses. I'm already familiar with some ML functions like Sigmoid, ReLu, and TanH; but I have not yet built my first artificial neural network. Personally, I enjoy Statistics more than Calculus, so I'm hoping that I don't need to take more Calc courses to be successful in this field. I've already taken Stats 1 which taught Standard Deviation, Variance, Distribution, and a few other concepts, but it didn't cover Linear Regression and my Calculus course hasn't covered Gradient Descent. If there are only a few more essential topics for the math used in ML, then I can learn on my own if I know what to study.

[babayaga]

This was better than many, but still its not good for beginners. I means lot of things are not told detail. I might sound totally stupid with below questions, but I am a beginner
Q1. How do you get this pink line at 2:55 ?
Q2. in Y=mx+b, how can X be independent ?
Q3. In this video where specifically you told about the prediction that was made by algorithm we develop, and how we compare it with the correct data ?

[jiyaadnaeem]

Absolutely beautiful explanation of where different disciplines come in to solve a real Machine Learning problem - especially since I was looking for an apartment around NYC :) . Thank you so much for this!

[mariamonica]

Siraj, I love your videos, I watch them with subtitles because I am learning English still😀, I intend to enter this area in the future. Congratulations 👏👏😊