When did you realize that machine learning is actually just math and has nothing to do with software """engineering"""?
It's just mathematics implemented in python. Go read an ML program. It's graduate level math (or lower) implemented in 1st year level code. Why do you think people teach ML in Matlab/Octave?
it's not even math. it's statistics.
this
> Why do you think people teach ML in Matlab
> Why do you think people teach Matlab in Matlab
seriously?
right away.
People teach machine learning in matlab, python and R because it is easier to implement mathematical stuff in these languages.
>machine learning
>its not learning as much as optimising
>It's just mathematics implemented in python
So what? Physics is just mathematics implemented in atoms. Chemistry is just atoms implemented in molecules. Biology is just molecules implemented in cells. Medicine is just cells implemented in tissues. Just because you are good at maths doesn't mean you are a neurosurgeon OP.
>Physics is just mathematics implemented in atoms.
idiot
Statistics is math you fucking idiot.
>It's just mathematics implemented in python
If you're looking at a shitty markov chain, maybe
Point and laugh at the retarded engineer.
Literally who makes an AI in python though
for that kind of processing power you cannot use such a high level language.
>Chemistry is just atoms implemented in molecules
idiot
It really is. I have no idea about programming (besides the basics) and still use machine learning sometimes. R packages makes them easy as hell
>seriously?
ML = machine learning
If you are talking about what people actually mean when they say machine learning or AI nowadays, which is deep learning, then it's just python.
Most of what people do is just training different architectures on a deep learning platform. And there's virtually no competing platforms which is not operated on python by now (a couple but they are getting phased out). At the end of the day software engineering is a waste of time for deep learning specialists and any possible gain in terms of performance over python implementations is vastly overshadowed by the development time loss. With deep learning, you basically want to change shit up all the time and to be able to do it easily on a high level language is key.
>ML = machine learning
ML is a Mercedes vehicle.
Almost all neural network research is done on python, with Theano. Python is the best option with best GPU support for ML.
> Matlab = machine learning
kys you're are self
I disagree.
You learn by using optimization.
If you watch infants, you see that they explore everything, how different objects feel, how they grasp different things to learn how to grasp other things.
Terms like machine learning and artificial intelligence is just terms.
We don't call it control theory, or optimization because it is different from that.
Wednesday afternoon, with my signal analysis course.
Build tool is software engineering, maths or stats could create models and analysis, but time you need make money, put in production you need software engineering.
openai are a joke and gym is a buggy piece of crap that literally, LITERALLY doesn't work.
Also not ML.