51:13 - Prof Boyd suggested to explore Julia 1:34:11 - PhD program is professor replication. To replace the number of professors that die in the given year. But train yourself broadly.
@csl13845 жыл бұрын
Boyd starts at 14:48
@konstantinburlachenko28436 жыл бұрын
59:00 - "Fun history about SpaceX and his S.Boyd former student who works at SpaceX" 01:16:00 - "Stephen about that today in areas which has interconnection between computer science and machine learning in fact has no separation between Theory and Practice nowadays. Separation of Theory and Practice is not so correct right now" 01:24:16 - "Stephen comment about apply SGD to non-convex objective like Neural Nets and various examples from history which still has theory/practice gap." 01:42:00 - "Stephen comment about convex optimization that it should be generic field for everybody from 15-20 years. Thing like Linear Algebra. And comment that apply Deep Neural Nets when we can solve problems exactly is a big mistake of sociey. There are various tradeoffs of generality and applicability, but whatever create End to End DL project is a big mistake, even my friends do it...." 01:43:45 - "Comments abotu Falcon 9 landing. You can use DL for Vision system and other very difficult problems, but not for problems that we know how to solve exactly, it has no sense to me. And I hope pretty soon people will start understand it."
@jimmylovesyouall Жыл бұрын
such a nice talk, thank you for uploading.
@alvarogustavotalaveralopez51044 жыл бұрын
Excelente presentación!
@romananalytics21823 жыл бұрын
He said same thing 2008 that Python is a real language and after a decade still sticking to it!!
@blakebaird1192 жыл бұрын
Don’t forget he published CVXR after doing this lecture
@michaelmellinger23242 жыл бұрын
@51:10 “Learn Julia” Sounds like he’s a fan
@Throwingness2 жыл бұрын
I've been watching for an hour. I have no idea how to spot, create, or solve a convex optimization.
@MrLee-gj2jz2 жыл бұрын
Do the full Stanford Course by Stephen Boyd; 10 weeks x 20 hours/week ...