Gilbert Strang: Singular Value Decomposition

[@lexfridman]

Full episode with Gilbert Strang (Nov 2019): kzbin.info/www/bejne/ona9gZmjfKh4oZI New clips channel (Lex Clips): kzbin.info Once it reaches 20,000 subscribers, I'll start posting the clips there instead. (more links below) For now, new full episodes are released once or twice a week and 1-2 new clips or a new non-podcast video is released on all other days. Podcast full episodes playlist: kzbin.info/aero/PLrAXtmErZgOdP_8GztsuKi9nrraNbKKp4 Podcasts clips playlist: kzbin.info/aero/PLrAXtmErZgOeciFP3CBCIEElOJeitOr41 Podcast website: lexfridman.com/ai Podcast on Apple Podcasts (iTunes): apple.co/2lwqZIr Podcast on Spotify: spoti.fi/2nEwCF8 Podcast RSS: lexfridman.com/category/ai/feed/

[@pycool7595]

Rotate, stretch, then rotate. Best one sentence explanation of SVD. Amazing, wow. I love professor Strang.

[@Actanonverba01]

AGREED

[@sauwurabh]

got a presentation about svd today. I'll be ending my presentation with this line.

[@ANunes06]

Prof. Strang: "I don't really bother with trying to visualize it. I just go straight to 10 dimensions and everything still works fine." Also Prof. Strang: "SVD breaks a Matrix into three pieces. A Rotation, a Stretch and then a Rotation." What a legend.

[@ozgegunaydin85]

i am a big fan of Gilbert Strang he tells us very very very clearly and pure! Thnxs

[@aiishg_]

He has true passion for teaching mathematics and that reflects in his videos. I have watche I don’t how many of his videos and it helped me a lot! What would have I done without his videos! 😅

[@Paul_Hanson]

I just ran into another KZbin video about linear algebra by one of Mr. Strang's pupils. I knew the name sounded familiar but it wasn't until just now that I realized he was the author of my linear algebra textbook in college (Linear Algebra and Its Applications). I really enjoyed his presentation of the subject and I think he really helped me appreciate it.

[@hyperbolicandivote]

Lex, thanks for bringing interesting universals to the internet.

[@muttleycrew]

Gilbert Strang is being modest, he discovered the beautifully simple Strang decomposition which writes any matrix, A, as a product of the so-called column space of A and the reduced row echelon form of A. Matrices aren't new, decompositions aren't new either, but it took Gil Strang to find that almost unbelievably simple relationship.

[@epicmarschmallow5049]

Everything I could find on the "Strang decomposition" was just CR decomposition, which was definitely not discovered by Strang

[@SigmaChuck]

As a math teacher, he is my idol.

[@rinkaghosh7961]

Thank you prof. Strang !

[@eriknovak496]

I've always really enjoyed thinking of matrices through the Jordan decomposition, besides the SVD too

[@sujitbasu2890]

He is a God of Linear Algebra Teaching. He changed my concepts on this subject.

[@3bdo3id]

He is so simply good grand 🧡

[@APaleDot]

Ooh, a rare chance to correct a renowned professor for a small inconsequential math mistake! I've got to jump on this! There are actually much _more_ than 10 ways to rotate an object in 10-dimensional space. Degrees of rotational freedom follow a combinatoric growth rate. So for instance, there are 6 ways to rotate in 4-dimensional space, 10 ways in 5-dimensional space, and so on...

[@thinkandmove479]

Thank you for clarification. I also just wondered, if he really made a small mistake here.

[@APaleDot]

@@thinkandmove479 I'm sure if he was actually working through a problem he wouldn't have made that mistake. It was just an off-the-cuff comment in the middle of a sentence, and I thought it would be a good chance to explain some math.

[@Actanonverba01]

Great Stuff! I love it the way he explains it.

[@axelnnz]

Amazing, thanks profesor Strang It's such a fundamental concept to everything ml and ai and still, it sound so banal

[@harisridharan8693]

Wonderful explanation Professor

[@chrischoir3594]

Roll, Pitch, and Yaw

[@scientifically5812]

The King of linear algebra!!!!

[@AbdoMohammed-jt5ye]

And calculus

[@taeukham623]

Gotta study singular value decomposition right now

[@eldyy9328]

@Lex Fridman Hey, Lex. You should bring Dr Harold G. White on the podcast to talk about NASA's Advanced Propulsion Physics Laboratory. It may spark more interest in the field.

[@lexfridman]

Great recommendation. I added him to the list. By the way, I try to read all recommendations for guests. Most are fascinating people. I love it! Even if I don't respond, please keep them coming. I'm likely to interview them eventually if you post it.

[@TheRealKGD]

Lex Fridman please interview Jitendra Malik, Bill Freeman, Raquel Urtasun, Vladlen Koltun, ... (see the computer vision trend 😀)

[@lupelicious822]

I took a math course in art school, taught by an adjunct from UC Berkeley, where linear algebra concepts were taught to us. I failed horribly at math in high school but I felt like I "got" numbers after that experience.

[@zavierbanerjee5171]

🙏🙏🙏

[@notgoodatmathmmm6185]

amazing

[@kevalan1042]

SVD should be called RSR

[@user-saint]

Rotate, Stretch , Rotate

[@humblesoul8685]

After Guass, gil strang is linear algebra's promoter

[@aanchaldogra9802]

I genuinely feel 3blue1brown would be huge fan of you. He seems to be carrying your legacy forward with latest technologies of course.

[@gzitterspiller]

not even close

[@aanchaldogra9802]

@@gzitterspiller what I meant was he is using the technologies efficiently to teach. Ofcourse he can't match him in any other field.

[@dirkmichaelis6055]

To me, the picture looks wrong or unusual, I am used to U being m x n and sigma being n x n. Sure, this is not a math class, however ...

[@LilyMyLolita]

Well, it's actually more common that U being m x m, sigma being m x n, Vt being n x n. See en.wikipedia.org/wiki/Singular_value_decomposition

[@amonal42]

4:19 - "up to ten dimensions you got 10 ways to turn". That is not true. You have N(N-1)/2 ways to turn in N dimensions.

[@leonardod248]

comment the proof

[@amonal42]

@@leonardod248 en.wikipedia.org/wiki/Orthogonal_group#As_algebraic_groups

[@gzitterspiller]

You are talking about the degrees of freedom of a rotation matrix... Gilbert was talking about the independant axis you can turn.

[@amonal42]

@@gzitterspiller I know no meaningful way to talk about a single axis of rotation in 10-dimensional space. There are 2 axis that rotate and 8 axis that are fixed in basic rotation.

[@bl1398]

Rotation is 2d is around a point, in 3D around a line, in 4d around a plane, I guess it isn’t just n

[@MoritaJunichiro]

😁. 😞my Hous is Japan abe gurp Electric or electromagneticEvry nhgat I'm attacked every night🇯🇵🤮🤮🤮abe 🤮🤮🤮