SVM (The Math) : Data Science Concepts

SVM (The Math) : Data Science Concepts

ritvikmath

3 года назад

95,295 Просмотров

Ссылки и html тэги не поддерживаются


Комментарии:

TheOilDoctor
TheOilDoctor - 12.10.2023 17:50

great, concise explanation !

Ответить
dcodsp_1803
dcodsp_1803 - 24.09.2023 21:18

Thanks for such brilliant explanation really appreciate your work!!

Ответить
Sejmou
Sejmou - 23.09.2023 22:30

In case you're also having trouble figuring out how we arrive at k=1/||w|| from k * (w*w/||w||) = 1:

remember that the dot product of any vector with itself is equal to its squared magnitude. Then, w*w can also be expressed as ||w||^2.
||w||^2/||w|| simplifies to just ||w||. Finally bring ||w|| to the other side by dividing the whole equation by ||w||, and you're done :)

if you also have trouble understanding why exactly the dot product of any vector with itself is equal to its squared magnitude it also helps to know that the magnitude of a vector is the square root of the sum of squares of its components and that sqrt(x) * sqrt(x) = x

I hope that somehow makes sense if you're struggling, surely took me a while to get that lol

Ответить
Mahesh Sonawane
Mahesh Sonawane - 05.09.2023 19:32

🌟Magnificient🌟I actually understood this loss function in by watching once. Very nice explanation of math. I saw lot of other lectures but you cant understand math without graphical visualization.

Ответить
pradeep svs
pradeep svs - 29.08.2023 08:01

@ritvikmath, why is the intercept (b) is negative? The equation of the line/plane/hyperplane should be w1x1 + w2x2 + w3x3 + b = 0, i.e. wx+b = 0 should the line equation. isn't?

Ответить
Luis Tupac
Luis Tupac - 28.08.2023 04:10

Dude thank you! now these equations don't feel like they were pulled out of thin air. and the best part is I can work them out too! I haven't done linear algebra in almost a decade so I got stuck on the ||w||/(w*w) part for a good bit but this pushed me to refresh some concepts and figure it out! Thank you

Ответить
Ashar
Ashar - 06.08.2023 21:20

What an amazing video bro. Keep going.

Ответить
Yonatan Dan
Yonatan Dan - 20.07.2023 14:03

phenomenal

Ответить
დუბროვნიკელი დამიენი.
დუბროვნიკელი დამიენი. - 10.07.2023 01:40

უფალი დაგლოცავს, შენზე თესლი არავინ არ არის ამეების ახსნაში

Ответить
Pedro Colangelo
Pedro Colangelo - 07.07.2023 21:45

Once again, ritvikmath being a lifesaver for me. If I understand the underlying math behind this concepts, it is because of him

Ответить
Ghaleb Al-Gobi
Ghaleb Al-Gobi - 07.07.2023 02:14

You explained this topic perfectly! Amazing!

Ответить
Suckock Shititties
Suckock Shititties - 05.07.2023 12:17

You are an amazing elucidator👍

Ответить
Kanishk soman
Kanishk soman - 30.06.2023 09:05

Hi Ritvik, you are a great teacher of stats, calculus and ML/DL!

I have one question regarding the equations. Why is the decision boundary equation W.X - b = 0? Shouldn't it be W.X + b = 0. I know the derivations and procedure to find the maximal margin is not affected but I don't understand -b. Please let me know if the sign is inconsequential. If it is, why is it? Thanks!

Ответить
Max Patrick Olivier MORIN
Max Patrick Olivier MORIN - 25.05.2023 01:24

Great video

Ответить
anish mohan
anish mohan - 15.05.2023 20:41

Thanks for this wonderful video.

I understand that the equation of blue dotted line (plane) is W.X+b =0 .

But how can we decide the other two lines. I mean how those can be W.X+b = +1 and W.X+b = -1.

And if they are, then the width is 2 right? how we can maximize it, it is fixed isnt it so?

I know I am talking nonsense :) I dont have anyone else to ask this :)

Thanks in advance!

Ответить
Arvind S
Arvind S - 09.05.2023 05:09

Great work

Ответить
zarbose
zarbose - 05.05.2023 01:53

Incredible video

Ответить
Manik Chadha
Manik Chadha - 21.04.2023 08:35

Starting mein kya kiya bhai tune?

Ответить
Liu Pingting
Liu Pingting - 11.04.2023 04:58

amazing video!

Ответить