Daily reminder if you consider yourself a programmer and you haven't worked on any form of CNN/RNN/GAN or any other...

'That's not programming the compiler does most of the work for you' -ASM sperglords

I'm just saying that it is about as much programming as SQL is.

>NOT running locally
Are you that poor that you need to rent someones server?
Can't you afford a private datacenter?
That's laughable.

I understand what you're saying but at the end of the day learning from generic topologies that are shoved down your throat by tutorials is the first step. The reason why this subset of the field is not talked about as much here is due to the fact that the underlying concepts, as well as true application of new ideas requires deep understanding of statistical methods and cutting edge research, most of the users here are just too new to programming or aren't intelligent enough in the first place.

needs more buzzwords

>tfw this is all generated by me, memebot 6.90

Attached: 9KGFr4k.png (600x600, 189K)

>assembly
>obsolete
God I hate zoomer programmers.

Machine learning is a meme, didn't some survey study conclude that 86% of medical studies using ML had irreproducible results when applying different datasets using the same training sets? Even facial recognition, probably the most mature application of ML today, doesn't work properly when used on asians and generates false positives all over the place.

>doesn't work properly when used on asians
yes! honest chinese citizens have nothing to worry about, only used for tracking foreign elements

Attached: 0bc726dc-c3f5-11e7-9f00-d8b0ccf89a9f_1280x720_130405.jpg (1200x800, 371K)

It obviously works when you use it ONLY on asians (aka using asians as training set and using asians as data set). The problems arise when you start mixing ethnicities (pun intended).

upvoted! rolf

>hurr durr i plug in data to tensorflow i am super smart because the word neural network confuses my mom

this

code monkeys:
LLVMErrorRef LLVMOrcCreateLazyCompileCallback(
LLVMOrcJITStackRef JITStack, LLVMOrcTargetAddress *RetAddr,
LLVMOrcLazyCompileCallbackFn Callback, void *CallbackCtx) {
OrcCBindingsStack &J = *unwrap(JITStack);
if (auto Addr = J.createLazyCompileCallback(Callback, CallbackCtx)) {
*RetAddr = *Addr;
return LLVMErrorSuccess;
} else
return wrap(Addr.takeError());
}

REAL programmers:
import MachineLearning

>like dude AI and ML isn't relevant bro

ML is not that difficult, friend. Look at githube - thousands of pajeets have implemented some shit in python, tensorflow, octave, whatever.

As long as shit is simple, the niche will be full (it is already full) and unless you have a proper math degree from a reputable school you are fucked.

I could train any kind of shit just by copypasting from github - but there is literally no demand. zero.

So, plain old math which is a formalized and systematized problem solving is still the king.

Really? What's the catch?