You have 10 seconds to justify yourself why you aren't studying machine learning

You have 10 seconds to justify yourself why you aren't studying machine learning

Other urls found in this thread:

developer.nvidia.com/devbox
twitter.com/SFWRedditImages

I am

I'm learning information theory. Machine Learning is a meme.

I'm waiting for the machines to actually learn it for me so I don't have to.

By then your entire existence will be pointless and you'll be marked for elimination by them.

I'm too stupid

You talking like my existence is not already pointless.

I'm too dumb to understand it

It's oblique

I got distracted by other stuff, I'll get right back on it OP

I tried but I failed to understand since it's not related to physics.

I'm not falling for memes

it's a crowded field mate

very hard to get in in a paid way

Probably will for my master thesis research project. Don't know shit about it yet so I figure it's a good topic to spend 7-8 months on. Seems useful as fuck for a truckload of different applications.

>not just getting in it as a hobby because its interesting and promising as fuck, not even saying that you can apply it everywhere, not just working in the field
You sound like a pajeet, I live in a third world shithole and i'm pretty excited about studying it in my free time while doing CS

But I am user. Currently studying a bachelor of computer science, majoring in Artificial Intelligence and Data Science

You sound like a child.

why study it when you can use pre trained models with easy to use frameworks to get standard tasks done ?

Because there's too much shit content on the web where I don't fucking learn anything.

I mean, I'm all for adding more tools in my analytical skill set, but I feel like most online stuff that pretends to teach it are lacking in optimization and statistics. Not to mention use cases.

Well, that and I haven't found a book I can understand at an appropriate level. It's always either too simple or too advanced. Or maybe I lack the patience to read shit for a hobby.

Basically these. Except information theory is a meme as well.

Because I am the CEO of a machine learning startup and I let the drones bother with that now

List topics that I need to learn to follow up with Machine Learning.

Literally 0 jobs in New Zealand

I'm trying to but I'm still too much of a pleb with the underlying math for a lot of it.

this

no idea where to start or if there is even any practical use (for me)

>not owning at least one deep learning work station

My brain works much faster than any machine ever will.

>SLI
>Deep Learning
lmao

>Disabled

>not server CPU with ECC RAM
You're fucking retarded.

>falling for the ECC meme
Only useful in simulation, retard

Yes, you've wasted all that money retard.

>implying
developer.nvidia.com/devbox

But I did take up statistics.

So? ECC is a must have.

what? where do you live? its the opposite in london

You don't need ECC RAM in deep learning. Most of the computation happens on GPU and training states (models) are not hard to backup.

Sure, tard.

Do you have any argument at all?

ECC CPU RAM is useless in deep learning, as computation happens in GPU.

GPUs with ECC RAM are often as well geared towards double-precision (64-bit) scientific simulation, which is useless in deep learning. Research has shown that as little as 13 bit is enough for training, and as little as 8 bit is enough for inference.

Mini-batch stochastic gradient descent, the most used deep learning training routine, is far from noise intolerant. Most deep learning frameworks offer ways to cap the gradient so even if a number is flipped to NaNs it wouldn't damage the training process. In fact, noise serves as a regularizer in training. Hell, one of the most used tensor operation, the reduction, isn't even deterministic anyways.

It's for niggers who can't into real computer science, like system architecture.

How do you get access to NVidia's deep learning API? Is it just a matter of registering on their website?

What rational reason do I have as a human being for wanting machines to think for themselves?

>guys why do we need computers to do tasks? we can do it ourselves!

>Do you have any argument at all?
Yes, my argument is that you're a tard. And you have no idea how computers work.

Why limit yourself to a subtopic of AI?

AI is a subtopic of machine learning, not the reverse

Yes. To download CuDNN or DIGITS (which is a shit-tier DL framework) you need to register. Registration needs to be manually approved but should usually be done within one work day. Just fill in you're a hobbyist etc. they don't judge.

because it's boring as fuck

>tard tard tard xDDDD
go back to \b\ user

I'm learning signal processing. Machine learning is a meme.

I need to finish something big first, if anyone helps, I'll start on reading

but it is though...matrix operations, linear algebra, etc. I was a physics major and it made understanding the theory way easier.

This is correct. Some of the top machine learning researchers and applied statisticians have backgrounds in statistical physics

>Deep Learning
>not using an advanced OS preferred for such applications (GNU/Linux)

I think you mean linux

GNU-distributions of Linux like debian GNU/Linux are notoriously unequipped driver-wise

Jupyter Notebook + Tensorflow runs hassle free under Windows 10. Kernel launching overhead is almost non-existent.

I'm going to study it.
Can anyone tell me what should i read, do etc?

>Tensorflow
>runs hassle free under Windows 10.
HAHAHAHAHAH
FUCKING RETARD

You don't even know what you're wasting

Because I already have a MS in AI.

Fuck off freedumb bitch

If you're using tensorflow under windows (which means under linux with an added layer), while caring enough about performance that you have multiple high end gpu's... you're a moron

>I have no idea what deep learning bottlenecks are