|
it's vector calculus and linear algebra all the way down the rest is figuring out how to approximate approximations to statistically mostly correct estimators of models that are vaguely inspired by hilariously simple neuroscience garbage
|
#
¿
Aug 21, 2017 02:15
|
|
|
|
| # ¿ May 22, 2024 07:06 |
|
|
lol no the book you want is called "deep learning"
|
|
#
¿
Aug 22, 2017 01:34
|
|
|
eschaton posted:except when it's symbol and list manipulation all the way down maybe the truth is in the middle
|
|
#
¿
Aug 23, 2017 18:07
|
|
|
JewKiller 3000 posted:yeah. math is hard. that's why the classes are like 20 weeks long each and reading doesn't really do poo poo for really understanding and being able to apply that stuff. you've got to actually do some exercises and whatnot. otherwise you're just going to be spending all your time being all "why are we doing this eigendecomposition thing here?" instead of "oh yeah if the eigenvalues are > 1 then iterations are obviously gonna diverge. duh"
|
|
#
¿
Aug 24, 2017 04:14
|
|