Тёмный

Lengths, angles, projection, correlation | Linear algebra episode 2 

All Angles
Подписаться 10 тыс.
Просмотров 2,6 тыс.
50% 1

#vectors #linearalgebra #dotproduct #innerproductspace #orthogonality #correlation #fourier_series
The dot product has many cool applications in artificial intelligence, audio engineering, and classical mechanics. It measures the correlation between different pieces of data, which allows AI algorithms to discover new concepts. It also measures lengths and distances so that we can find out how closely one sound wave approximates another. We also look at a few weird and exotic inner products such as the taxicab distance.
Please support our channel on Patreon, and get early access to new videos: www.patreon.com/user?u=86649007
Here are many extra links you can follow if you want to dig deeper:
[WIKI 1] en.wikipedia.org/wiki/Dot_pro...
Here you can find the higher-dimensional proof for the equivalence between the two formulas for the dot product.
[3B1B 1] • Dot products and duali...
This video explains the dot product using the concept of dual vectors. We will talk about those later when we cover tensor algebra.
[ZS 1] • The real world applica...
Some cool real-world applications of inner products. One of them tries to find similarities between documents, by encoding all words as vectors first.
[TB 1] • The geometric view on ...
Derives the orthogonal projection formula. Then applies it to the standard basis vectors, to show that it retrieves the coordinates of a vector.
[TB 2] • Least Squares Approxim...
• Reducing the Least Squ...
Uses dot products to find the closest solution to a linear equation.
[MTB 1] • Linear Algebra 20L: Th...
Expressing the dot product in a non-orthogonal basis. This leads to the very important concept of a metric, which we will encounter again in future videos on All Angles.
[MTB 2] • Introducing the Inner ...
This is a very good in-depth series dedicated to inner products. It tackles many related topics such as symmetric matrices, metrics, Fourier analysis, Legendre polynomials, Gaussian integration techniques, and much more. Pavel Grinfeld is one of the best teachers on the web.
0:00 Why do we need the dot product?
2:15 Some exotic examples of inner products
3:49 Definition of the dot product
5:12 Length/norm/modulus
8:42 Measuring angles
9:48 Dot product not always possible
14:04 Orthogonality & correlation
20:40 You cannot divide vectors (yet)
21:53 Projections
23:48 The projection formula
25:37 Using projections for approximations
26:34 Applications of projections in music & physics
28:17 Proof that the two formulas are equal
29:48 Rules for inner products
This video is published under a CC Attribution license
( creativecommons.org/licenses/... )

Опубликовано:

 

1 июл 2024

Поделиться:

Ссылка:

Скачать:

Готовим ссылку...

Добавить в:

Мой плейлист
Посмотреть позже
Комментарии : 25   
@alegian7934
@alegian7934 7 месяцев назад
the apples and oranges argument for dot products is really neat! I'm the type of guy that immediately looks for rigor - and that just clicked really well :)
@alegian7934
@alegian7934 7 месяцев назад
also I wanted to note, I think you've found a really nice spot with the colors. Minimal, clean, contrasty
@AllAnglesMath
@AllAnglesMath 7 месяцев назад
Good, that's exactly what we were aiming for. It's always a little bit subjective. Glad you like it!
@martinsanchez-hw4fi
@martinsanchez-hw4fi 2 месяца назад
Great video. The inner product int always non negative. Just for vectors with themselves.
@TrollFunMineMafia
@TrollFunMineMafia 7 месяцев назад
I'm amazed how intuitively you explain these concepts, I remember struggling so much with these, thank you! One thing I always wondered, when I learned about vectors it was oftentimes shown with both being the same size for the calculations. However, me coming from a cs background, it made more sense that one should avoid storing every possible value in memory. The example, with the movie ratings, reminded me of this. So I've always wondered how one can still apply this math even tho every user might have a different sized vector, instead of assuming a rating of 0 if unrated.
@ccaiobianchi
@ccaiobianchi 6 месяцев назад
Amazing. Beautiful and well-explained. Keep going.
@scalex1882
@scalex1882 7 месяцев назад
Absolutely amazing content! Thanks for making these videos
@AllAnglesMath
@AllAnglesMath 7 месяцев назад
Thank you for the thank you ;-)
@DeathSugar
@DeathSugar 7 месяцев назад
Interestingly almost the most obvious applications of dot-product is missing - rendering things on screen, especially games. To make CPU or GPU to do things efficiently and out of millions poligons on the scene choose only those who is visible - dot product is indespensable. You do dot product between camera plane and target normal of the target polygon and you receive the number - if value of dot product is positive - than you see it and you should render it. It also known as frustum culling.
@AllAnglesMath
@AllAnglesMath 7 месяцев назад
That's definitely a cool application.
@badermuteb4552
@badermuteb4552 7 месяцев назад
you videos are the best on youtube, thank you so much.
@AllAnglesMath
@AllAnglesMath 7 месяцев назад
Wow, thanks! Can I quote you on my resume? 😉
@linuxp00
@linuxp00 7 месяцев назад
In specific settings where bases are given by a set of matrices, inversion is sometimes possible though, that the power of (multi)linear algebra. Congratulation for your excellent multidisciplinary lessons. P.S.: I'm missing chapters separations this video as they are very helpful to jump back and forth
@AllAnglesMath
@AllAnglesMath 7 месяцев назад
Thanks! We'll see what we can do about those chapters. It's a valid point, I'm just not sure how to configure that in RU-vid.
@AllAnglesMath
@AllAnglesMath 6 месяцев назад
Hi @linuxp00, It turns out that adding chapters to RU-vid videos is very easy. So I added them to the two linear algebra videos already. We would appreciate your feedback: Did we separate the chapters in a useful way? Do you think we need more of them, or less of them? Are the titles OK? Thanks for helping us make the videos more useful for everybody!
@linuxp00
@linuxp00 6 месяцев назад
​@@AllAnglesMath I believe this one now has a very neat set of chapters, which makes it much easier to follow along with the video, especially with my ADHD's short attention span. Hahaha. 😅
@NicolasMiari
@NicolasMiari 5 месяцев назад
16:26 lol they both agree on The Last Jedi
@carlostrebbau2516
@carlostrebbau2516 3 месяца назад
'All Angles' is right
@kasugaryuichi9767
@kasugaryuichi9767 7 месяцев назад
@alanthayer8797
@alanthayer8797 7 месяцев назад
U SHOULD LEARN GEOMETRIC ALGEBRA not this MAINSTREAM MATH! Dot products , conjugates & EVERY other math subjects is within GEOMETRIC ALGEBRA & GEOMETRIC CALCULUS! Goto Bi-Vector & SUDGYLACMOE RU-vid Channel fa Visuals!
@samueldeandrade8535
@samueldeandrade8535 7 месяцев назад
Wtf?
@linuxp00
@linuxp00 7 месяцев назад
Wow, man. Chill out a bit! He'll get there eventually, that is just a foundations for things to come.
@alanthayer8797
@alanthayer8797 7 месяцев назад
@@linuxp00 no disrespect ta him or da audience! Was tryna Expedite his LEARNING CURVE about GEOMETRIC algebra to Channels that's ONLY Specializing in it alone bcuz ALL othe maths r Derivatives of GA! Which is y NO SCHOOLS introduces it
@alanthayer8797
@alanthayer8797 7 месяцев назад
@@samueldeandrade8535 no disrespect ta him or da audience! Was tryna Expedite his LEARNING CURVE about GEOMETRIC algebra to Channels that's ONLY Specializing in it alone bcuz ALL othe maths r Derivatives of GA! Which is y NO SCHOOLS introduces it
@samueldeandrade8535
@samueldeandrade8535 7 месяцев назад
@@alanthayer8797 look for a psychiatrist. You have serious problems.
Далее