Twitter | Pretraživanje | |
QC 27. stu
mathematicians mostly don't work on anything remotely useful and they rely on nobody else understanding what they do to keep this up; schools tolerate this b/c prestige + somebody needs to teach and TA calculus
Reply Retweet Označi sa "sviđa mi se"
QC 27. stu
Odgovor korisniku/ci @QiaochuYuan
the standard counterargument "but you never know what might be useful in 50 years!" ignores opportunity costs. we're allowing some of the smartest people around to nerdsnipe themselves and each other into spending all their time on beautiful useless shit
Reply Retweet Označi sa "sviđa mi se"
QC
a bigshot professor at berkeley once tried to dunk on ML by saying that it basically only relied on 18th century math; imo this is a much bigger dunk on math
Reply Retweet Označi sa "sviđa mi se" More
QC 27. stu
Odgovor korisniku/ci @QiaochuYuan
the other standard counterargument "but muh beauty of math!" - look. we are already drowning in beautiful math. we have a disgustingly beautiful hoard of it already, more than we could ever need. i'm talking full RPG inventory of megalixirs and all that other shit you never use
Reply Retweet Označi sa "sviđa mi se"
QC 27. stu
Odgovor korisniku/ci @QiaochuYuan
actually it's more like an RPG with a crafting system where you can craft really powerful weapons and armor that you don't yet have the stats to equip, and instead of leveling or figuring out what you really need to craft you just keep crafting more and more powerful stuff
Reply Retweet Označi sa "sviđa mi se"
QC 27. stu
Odgovor korisniku/ci @danlistensto
where 👏 is 👏 the 👏 lie
Reply Retweet Označi sa "sviđa mi se"
Toph 27. stu
Odgovor korisniku/ci @QiaochuYuan
Oh wow. Yes
Reply Retweet Označi sa "sviđa mi se"
AdA 6. pro
Odgovor korisniku/ci @QiaochuYuan
If we ever form wholistic theories of ML the will almost certainly invoice a lot of information geometry, largely developed in the past 3 decades.
Reply Retweet Označi sa "sviđa mi se"
AdA 6. pro
Odgovor korisniku/ci @QiaochuYuan
Wow. The typos... 🤦
Reply Retweet Označi sa "sviđa mi se"
Ty 28. stu
Odgovor korisniku/ci @QiaochuYuan
The Next Generation operator for finding R0 values for mathematical epidemiology was developed in the 90s. Wavelet analysis took off in the last few decades too, with the work of Ingrid Daubches.
Reply Retweet Označi sa "sviđa mi se"
Daniel Klein 28. stu
Odgovor korisniku/ci @QiaochuYuan
It’s not Marx or Lenin’s fault that linear programming was hardly developed and some big fixed point theorems were unknown when they wrote.
Reply Retweet Označi sa "sviđa mi se"
craigsuperstar 28. stu
Odgovor korisniku/ci @QiaochuYuan @jimmyvooh
This is "you're awesome!"ing on math, not dunking.
Reply Retweet Označi sa "sviđa mi se"