Machine learning without pythonn

>wants to do machine learning
>hates python
user, wat do

Other urls found in this thread:

caza.la/synaptic/#/
wagenaartje.github.io/neataptic/
github.com/josephmisiti/awesome-machine-learning
junolab.org
twitter.com/NSFWRedditGif

Sort of in the same boat, but after playing around with a lot of the gpu training libraries, I realized that a lot of the times it just backend shit that's in python. A lot of the time you don't have to write scripts and stuff in python necessarily.

Matlab

Same as you, probably will learn some lisp and prolog.

just suck it up and use python

>hurr durr, I wanna make website but I hate HTML and Javascript!

that's what you sound like

There are ML frameworks in Lua and Javascript, and most of tensorflow and numpy is written in C. And though it's losing traction, R is still pretty big esp. among the academics.

Caffe has a C++ API

>There are ML frameworks in Lua and Javascript
JavaScript? o.O

caza.la/synaptic/#/
wagenaartje.github.io/neataptic/

this, just get over it, python is so easy to use and there are so many libraries that you can do a lot with it, trying to do it any other way is gonna be more of a headache than it's probably worth. why NOT python?

>this, just get over it, python is so easy to use and there are so many libraries that you can do a lot with it, trying to do it any other way is gonna be more of a headache than it's probably worth. why NOT python?
It is the antithesis of functional programming?

>machine learning
>javascript

Obviously do it with Scala. Pretty obvious since some of the biggest ML frameworks like Spark's MLib are Scala anyhow. Plus ScalaNLP and friends. And more.

github.com/josephmisiti/awesome-machine-learning

Pick the language of your choice

PS: Obviously machine learning also benefits from the related, huge ecosystem around big data / big processing that is strongly on Java & Scala.

If you need to process giant data sets to train your software or do searches in that giant data set, nothing really will work better overall, or have more mature frameworks.

You could port existing python code to whatever language. That would be useful.

use Scala, but keep in mind Python already has a LOT of stuff done/adapted for it

> Python already has a LOT of stuff done/adapted for it
Eh, well, so does Java and actually Scala itself.

Apart from that, ultimately you can run a hybrid of all of these if it's opportune.

Matlab and R.

Julia is designed to replace Matlab, R, Python, Fortran and does so very well

Who the fuck cares?

>Who the fuck cares?
Anyone who does serious ML programming is not going to use Python babby libraries which are just black box wrappers over Fortran and C++ libraries. The fact that Python is slow and cant pass around high order functions is also a good reason not to use Python.

>Julia is designed to replace Matlab, R, Python, Fortran and does so very well
Does it have a chance in hell of catching on though?

So, there is no way around python for machine learning but python also sucks dick. Alright, makes no sense..

Is there a way to use Julia notebooks without a shitty web interface?

>Does it have a chance in hell of catching on though?
with Matlab and R you are stuck doing matrix math and only matrix math

with Python you cant pass high order functions and have lame ass vector handling

C++....muh templates

Fortran is ancient

Julia is blazing fast, simple syntax similar to Ruby/Javascript/Python, high order functions dont get cock blocked by OO scoping, and a full set of high order functions to work over container classes

unfortunately Jupiter notebooks is implemented in iPython, I would like to see it entirely rewritten in Julia

>machine learning

It's just a ridiculous fad and buzzword that gives millennial retards a hardon. I'd ignore it if I were you.

exactly "muh templates" are the power of C++
I don't understand how what you wrote is supposed to undermine it

ML with assembly.

This, or FPGA

i create all my ML/DL algorithms in pure assembly tyvm

> the state of Sup Forums

MATLAB

what is the state of julia?

junolab.org Or plugin visual studio code.

>built on Atom
I said without a shitty web interface

No. "Notebooks" live in web browsers.

Doesn't even matter if it's Jupyter, Zeppelin, Beaker or anything else (with all the languages they support), "Notebook" is web browser based.

You're programming boring old software in an IDE if you want a boring old IDE rather than a hip new notebook.

what exactly is the issue with python?

Then use R. Or use matlab if you want to never have a job.

>learn python
or
>understand that you can stop being a lazy asshole and simply start programming your own AI in pretty much any other language, there is literally nothing stopping you except access to gay ass mini-projects by jewgle faggots, hell, if you understand code itself you could probably understand what algorithms are doing with minimal googling as you could just look at source code and do your own shit. Now go out there and made a genetic algorithm from scratch you lazy fuck!
Make your choice.

Python