I highly recommend anyone who doesn't have a solid understanding of calculus check out 3blue1brown "the essence of calculus" series. It's relatively short, not boring, and does a really good job of teaching it. I also recommend his "essence of linear algebra" for whenever siraj makes a video on it.
Siraj, I am really glad you decided to go towards this slower teaching route. I think this will turn to to be amazing video series, looking forward to it. Some feedback on video: 1) At 5:34 it would be nice if you followed up the math with a example. Example could be suppose we want to do hand digit recognition, then X = pixels of 20x20 image in grayscale, Y = number in image, Error Function = formula for logistic error. 2) At 9:47 you show the python code for the partial derivatives of SSE. Before getting into the code, you could show how to compute dSSE/db and dSSE/dm. You should also put the gradient descent formula that you have at 0:19 and talk about learning rate parameter.
I love these new style videos - less random memes, more focus and relevance to the content, and better-placed humor. It really improves my learning. Thanks for your hard work!
man! I have taken almost more than 20 courses in machine learning. I teach these courses at university too. I have never seen such a good explanation. man you made me your disciple
Siraj! you are the most awesome guy.You made me fall in love with machine learning, even just being in 3th term in my university. Because of you, I want to work in this area after completing my studies. Thank you for that so much! Keep up the work!
Albert Einstien, George Bernard Shaw, and Jane Goodall represented humility with intelligence and all abstained from eating and destroying their intelligent coexisting beings. The coming hyper intelligence must be endowed with these characteristics for us to coexist, survive, thrive, and evolve. Thank you for bringing awareness and hope.
This is immensely helpful @Siraj. I have spent so much time researching the math on various ML algorithms and it is frustrating sometimes as one concept requires you to know a bunch of other things and then I spend time on them and it's like a never ending cycle. It's good to have consolidated content which relates everything to the main goal, which is implementation in ML.
I'm going into some courses that require me to know the math of ML next semester and this is perfectly timed. Looking forward to it! Also, your jacket is dope.
Hey Siraj, you have always been one of my all time favourites! You are great! I am a machine learning enthusiast and making my way to greater things. I am pursuing my undergraduate studies in India, and would hope to meet you someday in the future. Every time I watch your videos, they inspire me to do new things. Thank you so much for sharing your knowledge with us all!
Thank you so much for this course Siraj. I always wanted to get started with machine learning but I was afraid of mathematics. Due to your course, I'll be able to learn machine learning. Thanks a lot and keep making this type of courses in future.
You are awesome Siraj, you gave me hope to solve much than i ever expected to do in programming, i will try to be a good student and pass this on to the ext generation...Thank you once more for sharing.
Hey bro lovely work.. For those who are complaining as how fast he's describing the maths....my suggestion is to first understand this topic from book or any other online source...then I am sure you are gonna love his vids... Thanks
I take a seat in the first line. Thanks for your videos, are amazing !!. I think your work is making some people "rethink" their own relationship with computing and maths. I don´t find books or good tutorials about maths focus on machine learning or IA in Spanish. Only "academic" and general books, but it´s very rare any kind of programming, computing with maths good stuff.
Thank you very much for that video! I appreciate the way you demystify ML and the concepts behind it. A much needed contribution about the sometimes ill founded discussions around AI
Thank You Sir, Great Thanks you are great. Loved all your videos and was waiting for this video series, will be watching the whole series within a hour or two of the upload. I'll watch all the other series completely after this one. I owe you mate.
Oh Man, This will be amazing, I'm so glad i found your channel. Maybe at the end of these videos I'll find some interesting topic for my master degree on chemical engineering.
Just started to watch your videos. Love the way you explain and make things interesting. Joined the wizards channel on slack. Learning regularly. This should've been our school classes. Thanks Siraj.
For linear regression you can also just use normal equations (which is more efficient) Just solve for when the gradient of the loss function is 0. for example for MSE the solution is: w = (transpose(X_train)* X_train)^(-1) * transpose(X_train) * y_train
I'm glad you based in Europe now, with your skills you will be very welcomed hope you enjoy your stay, it cold (winter) at the moment. You need to seriously considering do a TED Talk.
11:13 I'm not super used to this channel (but you absolutely got me there) but I'm pretty sure you don't need to memorize the power rule any more. I mean, that felt obvious from the start (i knew machine learning was about partial derivatives) and when you pondered which of the curves is easiest to optimize. I'm probably on my way to binge-watching this channel. Great quality video. I liked the art (have I recognized Van Gogh ' style?) in the background, which captures chaos in an almost magical way, which was quite on the point, I have to admit.
Hi Siraj. Thanks for your dedication to teaching, efforts to make crucial knowledge accessible and exciting, and patience :) After I've built some necessary foundations, I'm planning to take the ML, DL and ultimately AI specialities on Udacity.
Thanks for starting this series Siraj. I've really been wanting to learn this machine learning stuff but I couldn't understand it because 10th grade math (actually 12th grade math because I started algebra in 7th grade) a doesn't cut it.
Woowowoooowwww sir!! Finally I find something that I or probably most of enthusiastic looking for or wanted to to learn!! And then it'll be great experience to learning from "YOU", really.... Thanks a looottt Sir 😇😇
I thought of adding subtitles to the video, but the auto-generated ones have become so accurate, I don't feel a need to do so. Machine learning is LIT.
Thanks Siraj ,this is AWESOME and so interesting. Come here because someone introduced you as an 'UA-cam celebrity of ML' , which sounds so weird ahahah (but you are). It's easy to make the command lines or click in the statistic program, but understand how it works is way more important than that
Really good Initiative Man! Bringing simple practice exercices to these therorical concepts help very well to get started for real! It's not obvious at all how to link those concepts to real world problems. This next three months will be great! Since you're in EU now... If you'll come to Paris let me know.. I'll offer the coffe (or the tea.. Depending on the results of the classification 😉)
Hello, Siraj. I started to translate it to Portuguese. I hope someone will review it soon, then it will be available. Thanks for this great new course!
To all the beautiful people here: the reason that he squared the error, is so that he would get a parabolic function as you can see at 9:00. Parabolic functions have a local minimum, which is what he wants to find. Linear functions, however, do not.
Tthank you Siraj for everything, you are great!! After these 3 months i would like to watch a series about BigData, because you have already talk about ML/DL/RL and now the optimizations , but nothing about how to build a good infrastructure, how to manage and how to fuel our ML algo. What do you think?
I highly recommend anyone who doesn't have a solid understanding of calculus check out 3blue1brown "the essence of calculus" series. It's relatively short, not boring, and does a really good job of teaching it. I also recommend his "essence of linear algebra" for whenever siraj makes a video on it.
agreed
Breh. Thanks. This helps a lot!
Thanks for the great info! It's amazing! Should I take more lectures like MIT calculus after complete 3blue1brown videos? or enough?
Finally a description for the math behind machine learning 👍🏻
It's time
I am a philosopher of science and not a coder but I seriously appreciate these videos. Very clear and well put together. Thank you for making!!!
Siraj, I am really glad you decided to go towards this slower teaching route. I think this will turn to to be amazing video series, looking forward to it.
Some feedback on video:
1) At 5:34 it would be nice if you followed up the math with a example. Example could be suppose we want to do hand digit recognition, then X = pixels of 20x20 image in grayscale, Y = number in image, Error Function = formula for logistic error.
2) At 9:47 you show the python code for the partial derivatives of SSE. Before getting into the code, you could show how to compute dSSE/db and dSSE/dm. You should also put the gradient descent formula that you have at 0:19 and talk about learning rate parameter.
Dude, you are a VERY good teacher. Honestly no BS, you're the best I've ever come across on the internet. Thank you!
I love these new style videos - less random memes, more focus and relevance to the content, and better-placed humor. It really improves my learning. Thanks for your hard work!
Couldn't have started this course in a better moment Siraj!!! I will start supporting you on Patreon, you deserve it! Keep on man! 💪
man! I have taken almost more than 20 courses in machine learning. I teach these courses at university too. I have never seen such a good explanation. man you made me your disciple
Had been searching for this kind of course since the last two months .Finalyyyyy.High expectations from this course.
Siraj! you are the most awesome guy.You made me fall in love with machine learning, even just being in 3th term in my university. Because of you, I want to work in this area after completing my studies. Thank you for that so much! Keep up the work!
thats so amazing! inspiring
Albert Einstien, George Bernard Shaw, and Jane Goodall represented humility with intelligence and all abstained from eating and destroying their intelligent coexisting beings. The coming hyper intelligence must be endowed with these characteristics for us to coexist, survive, thrive, and evolve. Thank you for bringing awareness and hope.
Nothing constructive to say. I'm blown away by the quality and quantity of your output.
This is immensely helpful @Siraj. I have spent so much time researching the math on various ML algorithms and it is frustrating sometimes as one concept requires you to know a bunch of other things and then I spend time on them and it's like a never ending cycle. It's good to have consolidated content which relates everything to the main goal, which is implementation in ML.
I'm going into some courses that require me to know the math of ML next semester and this is perfectly timed. Looking forward to it! Also, your jacket is dope.
Hey Siraj, you have always been one of my all time favourites! You are great! I am a machine learning enthusiast and making my way to greater things. I am pursuing my undergraduate studies in India, and would hope to meet you someday in the future. Every time I watch your videos, they inspire me to do new things. Thank you so much for sharing your knowledge with us all!
Thank you for this Siraj! Can't wait to see what comes next!
Ps those functions look beautiful in Python 😍
thanks Daniel!
Thank you so much for this course Siraj.
I always wanted to get started with machine learning but I was afraid of mathematics. Due to your course, I'll be able to learn machine learning. Thanks a lot and keep making this type of courses in future.
You are awesome Siraj, you gave me hope to solve much than i ever expected to do in programming, i will try to be a good student and pass this on to the ext generation...Thank you once more for sharing.
Hey bro lovely work..
For those who are complaining as how fast he's describing the maths....my suggestion is to first understand this topic from book or any other online source...then I am sure you are gonna love his vids...
Thanks
I love that you threw in the Part play Brian Crain on the piano, one of my favorite modern day pianist
This will be a groundbreaking series i'm so looking forward to it !!!
u called it
I take a seat in the first line. Thanks for your videos, are amazing !!. I think your work is making some people "rethink" their own relationship with computing and maths.
I don´t find books or good tutorials about maths focus on machine learning or IA in Spanish. Only "academic" and general books, but it´s very rare any kind of programming, computing with maths good stuff.
Thank you very much for that video! I appreciate the way you demystify ML and the concepts behind it. A much needed contribution about the sometimes ill founded discussions around AI
You the greatest bro. Made me comment on a an educational video for the first time. How did not get to know this channel earlier
Thank You Sir, Great Thanks you are great. Loved all your videos and was waiting for this video series, will be watching the whole series within a hour or two of the upload. I'll watch all the other series completely after this one.
I owe you mate.
Oh Man, This will be amazing, I'm so glad i found your channel. Maybe at the end of these videos I'll find some interesting topic for my master degree on chemical engineering.
Siraj Thank you so much for this! This is the best explanation of ML maths I've seen so far. You're totally awesome :)
Just started to watch your videos. Love the way you explain and make things interesting. Joined the wizards channel on slack. Learning regularly. This should've been our school classes. Thanks Siraj.
For linear regression you can also just use normal equations (which is more efficient)
Just solve for when the gradient of the loss function is 0.
for example for MSE the solution is:
w = (transpose(X_train)* X_train)^(-1) * transpose(X_train) * y_train
Even if I don't really like math, I will definitly try to learn to do this. It looks like magic !
I'm glad you based in Europe now, with your skills you will be very welcomed hope you enjoy your stay, it cold (winter) at the moment. You need to seriously considering do a TED Talk.
Wait for it, THIS IS GONNA BE LEGENDARY!!!
Wow what a magnificent change in the pace.
Congratulations Siraj, I really love the work you do.
I follow you since your 5k subscribers and you provide such a great contents that you deserve your huge growth here ;)
you had me at "we're not going to be using a framework" 😍
one word ... AMAZING I LEARNED MORE HERE THAN IN 6 TH GRADE
11:13
I'm not super used to this channel (but you absolutely got me there) but I'm pretty sure you don't need to memorize the power rule any more.
I mean, that felt obvious from the start (i knew machine learning was about partial derivatives) and when you pondered which of the curves is easiest to optimize.
I'm probably on my way to binge-watching this channel.
Great quality video. I liked the art (have I recognized Van Gogh ' style?) in the background, which captures chaos in an almost magical way, which was quite on the point, I have to admit.
Hi Siraj. Thanks for your dedication to teaching, efforts to make crucial knowledge accessible and exciting, and patience :) After I've built some necessary foundations, I'm planning to take the ML, DL and ultimately AI specialities on Udacity.
been waiting for a series like this ever since i found the channel!!!!!
i know
Finally off style! Less distractive and slowed the heck down.. Thank you, I say!
I feel so satisfied that I finally understand this... onto the next video
I need more videos on The Math of Intelligence! Awesome job!
best video yet, definitely continue this series
You are doing a WONDERFUL job, bro. Keep it up. I do owe you a lot !!!
Cheers!
Finally... someone who will explain the maths...
Yes! Thank you Siraj
Awesome! Looking forward to the course!
OMG. Finally it's happening. Much needed. You're amazing.
Thanks for starting this series Siraj. I've really been wanting to learn this machine learning stuff but I couldn't understand it because 10th grade math (actually 12th grade math because I started algebra in 7th grade) a doesn't cut it.
I am very excited for this series, thanks.
amazing video @Siraj, can't wait for the next one
I am waiting for Every Single Video in This Series to Change the Planet and Human race forever.
you called it
Woowowoooowwww sir!!
Finally I find something that I or probably most of enthusiastic looking for or wanted to to learn!! And then it'll be great experience to learning from "YOU", really.... Thanks a looottt Sir 😇😇
I've only watched first 29 seconds I stopped and came to say You're awesome and I owe you :')
dhrumil barot same here tbh :D
+1
np love u
Is that the same as "p love you" ? ;)
np hard love
Another outstanding video. I can't believe I don't have to pay for these. THANKS!!
that intro though is the best i've ever seen.
I thought of adding subtitles to the video, but the auto-generated ones have become so accurate, I don't feel a need to do so. Machine learning is LIT.
It's just goes over my head.
that's what i was waiting for,,, your reaaally geting better thumbs up for that
You're doing so much good to the world man. Thanks for the awesomeness, huge fan here
thx
That was real hype! And it was for a good reason! Go Math of Intelligence!!
I agree with other comments, Thank you!, I need a better understanding of the math behind the ML
That's really a helpful topic. Thanks Siraj. You are awesome.
thx
I have been waiting! Siraj you are dope. You nailed it again 😁😁😁
Can't wait to see next videos. really needed that so much thank you @SirajRaval
Just did a project where I had to use gradient decent to find the max kurtosis for blind sound source separation using independent component analysis.
Finally! Looking forward to see CNN, LSTM and other NN architectures in this series as well :) Ty
Thanks Siraj for another great video, really looking forward to this series!!!
Man my love neurons activate hard for you, thanks for these videos. You are not the most scientific, but you are most human ML teacher (that I know)!
Thank you so much Siraj. You are making a huge difference :)
Super excited!
Siraj thanks very much to share all this knowledge
Great introduction, as well as the resource links in the description.
thx
Thanks Siraj ,this is AWESOME and so interesting. Come here because someone introduced you as an 'UA-cam celebrity of ML' , which sounds so weird ahahah (but you are). It's easy to make the command lines or click in the statistic program, but understand how it works is way more important than that
thanks Dongqi!
Thank you!!! Math has been the one thing holding me back.
Great lesson, I had to study for Machine Learning exam from Bishop.I'll keep an eye on your courses they are to much fun to whatch
I really love your channel. Best one on topic of ML/AI
You are an absolute legend, much thanks from Australia :)
thanks Samich!
dude your output is massive! i want to watch all your videos so can you please chill out for a couple of weeks?! thx!
Yay! Exciting!
Thanks Siraj, great video as always!
Really good Initiative Man!
Bringing simple practice exercices to these therorical concepts help very well to get started for real!
It's not obvious at all how to link those concepts to real world problems.
This next three months will be great!
Since you're in EU now... If you'll come to Paris let me know.. I'll offer the coffe (or the tea.. Depending on the results of the classification 😉)
thanks Akai! will let u know!
You're awesome, Siraj !
YES DUDE THIS IS WHAT I NEED RIGHT NOW!!!
awesome
Finally, I've been waiting for this, This is awesome, Now do a Video on Designing a Neural Net Architecture...
cool
That's awesome dawg, much needed. Keep up good work!
PS Sorry if dawg is inappropriate, just wanted to use it in the sentence.
YESSSS I'VE BEEN WAITING SO LONG FOR THIS
thanks so much for your wonderful work siraj!
Hello, Siraj. I started to translate it to Portuguese. I hope someone will review it soon, then it will be available.
Thanks for this great new course!
i will thx
[6:35] Code begins (appreciate all the other fun too, Siraj)
thx
The best series!
To all the beautiful people here: the reason that he squared the error, is so that he would get a parabolic function as you can see at 9:00. Parabolic functions have a local minimum, which is what he wants to find. Linear functions, however, do not.
thanks Max
hmm parabolic function have minimum and also continuously differentiable which makes finding gradient easy
Sirajjjjjj you are the best man!
This is awsm Siraj.. No Words to explain ur great work . Keep going
Great initiative, Siraj!
Siraj this is amazing (and so is the timing) Thanks
Siraj, thank you so much for your contents!
Here is my implementation of the challenge! Hope yo like it!
github.com/EricAlcaide/Math_of_Intelligene/tree/master/Gradient_Descent
great work Eric, loved the focus on diamonds, well explained this was a very close challenge. keep it up!
Tthank you Siraj for everything, you are great!!
After these 3 months i would like to watch a series about BigData, because you have already talk about ML/DL/RL and now the optimizations , but nothing about how to build a good infrastructure, how to manage and how to fuel our ML algo. What do you think?