What do you think the next biggest field in programming will

>>8653799
Programming good,high performance, robust and secure software tailored to ones needs

because deep "learning" bubble will bust sooner or later

>>8653799
>machine learning
>field in "programming"
brainlet pls get off this board
differential programming will be the next big thing. you heard it here first.
>>8653950
deep learning bubble will not burst in the near future. it has just started and people who jump into the field in the next 10-20 will be guaranteed good money. every company will need a machine learning "specialists" or some kind to be competitive.

>>8653973
actually, it's differentiable programming. i made a typo

>>8653799
What's machine learning? I'm gonna start mechE. What can do to get an advantage and learn this?

>>8654034
Machine learning is using a program to learn from a set of data, to form a meaningful conclusion from said data, and maybe extrapolate further data.
You will probably not learn any machine learning in mechanical engineering

t. MechE grad student

>>8653973
>Implying that isn't already the case.

File: 1485484997103.jpg (105KB, 400x562px) Image search: [Google]

105KB, 400x562px

>>8654034

no offense mechie, but this requires both extensive programming and algorithm knowledge

things that an EE like myself posses

>>8654034
>>8654078
lol majors

>>8654078
>>8654070
Oh ok. No problem though I like programming just as a hobby

Anyway what can I do to maximise mechE gains? What I've learned so far is
>Get internship in first and second year breaks
> get a research assistantship leading to a publication in your 3rd year and finish with a good capstone that is relevant to research at a top uni

>>8653799

It's already the big field, so this prediction isn't really a prediction.

File: 1485963075729.gif (981KB, 200x200px) Image search: [Google]

981KB, 200x200px

>>8654078
>extensive programming and algorithm knowledge
>things that an EE possesses

>>8653799
>My guess is machine learning.

This shows you have no hope in humanity. Do you know what machine learning even is?

If solving problems ever becomes just bruteforcing it through machine learning then that's it for human advancement. No new revolutionary algorithms, no more breakthroughs in computational science. Just everyone using the same machine learning techniques to bruteforce even the simplest shit.

Not gonna happen. Machine Learning will stay a niche subject used only on very complicated stuff that will then get replaced by human made algorithms once we have studied the problems more.

>>8653799

If you want a prediction, look at the brain and then look at technology.

We already have machine learning, so why is technology not as powerful as the brain?

That's your answer. I'm not going to spell it out for you.

>>8654228
Thanks Ken M.

>>8654219
ML algorithms aren't "bruteforce" you retard.

>>8654219
>bruteforce
you are probably talking about some social science, bio, chemistry or physics retards trying to apply machine learning algorithm in their field to prove that they are edge and smart.

>>8654596

No problem!

Farming was the stomach.
Industrialization was the muscle.
Infrastructure was the blood and veins.
Information is the brain.

But we still have a stupid brain. We should fix that.

Regression models are racist because they imply that correlation means causation, which is a fallacy used to repress the black community by citing crime and IQ statistics.

>>8654647

I swear you are just this 1 nut job that every once in a while likes to spam /sci/ with shit.

>>8654652
That's just Ken M again

>>8653799
The next largest thing in CS would definitely be within the realm of AI/Machine Learning and the tangibility of grafting consciousness to an AI.

Though if you just meant in the incoming work force, you are correct that it is machine learning.

>>8654608
>ML algorithms aren't "bruteforce" you retard
how assigning random weights many times until one gets desired isn't bruteforce?

>>8654219
I think machine learning/neural networks is essential with respect to making an eventual strong AI.

I'm not quite sure how I feel about our eventual robot overlords.

Next biggest field in programming, prediction. big data based predictions.

>>8655241
It's painfully obvious that you don't know what you're talking about.

We search the function space by various optimization techniques. It's a directed search, not brute force at all....

>>8654608
>ML algorithms aren't "bruteforce" you retard.

Yes they are.

The thing is that we apply ML to problems that have no other solution, so we think that we are being smart and efficient but in reality that's not the case.

Out there, in the universe, there is an specific algorithm for all the crap we are using ML on. Just a perfect program that doesn't need all the bloat the usual ML system requires to even work.

If you don't believe me then use machine learning to make an AI that plays tic tac toe. Or use ML to do anything that already has a nice and clever solution.

>>8657030
This is idiotic. Machine Learning isn't applied to problems just because they are "hard" or have no other solution. Machine Learning is for problems that are dependent on extracting information from large volumes of data, like recognizing a particular person's face. Do you intend to write an algorithm for this?

>>8657030
No, again. ML algorithms are used to minimize a loss function defined on the problem. That's it. It's not brute force in any possible interpretation except an uninformed one.

For something as simple as a game you can easily implement a heuristic informed search like A*. You can also formulate it as a ML problem, but I'm not sure if the results would be great depending on the game.

>>8657030
Also, it *may* be the case that there is an "algorithm" for the problems we formulate as machine learning. This is a central idea of learning theory. It's this function or distribution which is the true reality of the data that we are trying to find using optimization techniques.

>>8653950
Don't you mean the "human programmer" bubble will soon burst?

>>8657044
>>8657058
This is a sort of narrow view of ML, ML is broader than just characterization / function optimization or computer vision. For example control problems are sometimes approached with ML (usually reinforcement learning or genetic algorithms), which I think are the kinds of problems where you would expect a clean non-MLey algorithm.

>>8657091
Genetic algorithms are not ML. They are a form of search.

ML and the theory of learning is based entirely on the idea of learning a distribution given a sample.

Given that ML has a lot of popular attention there may be misguided attempts to apply the techniques. It's also quite possible that you don't understand the nuances that the implementer sees in the specific problem which leads to a ML approach.

functional programming will rise

the javascript code monkeys will disappear into obscurity because they are incapable idiots

programmers will become involved with mathematics from type theory inward, and we will see a rise in algebraists who learned category theory before calculus

topologists will finally get what's coming to them

the world will be a better place

>>8655241
Most ML problems would be intractable by brute force methods. They would take longer than the lifetime of the universe to compute. Yet we have models that we can train in a few hours, or maybe days. Clearly they are not brute forcing.

>>8657320
this. finding the correct parameters to train a deep network is a pain in the ass but ML scientists developed many clever tricks to solve intractable problems of optimizing millions variable at once.

>>8654034
http://lmgtfy.com/?q=machine+learning

faggot

>>8654646
what's the fat?

probably drugs

>Machine learning is bruteforcing
>Statistic is now bruteforcing
Just to point out how stupid some people are. And you idiots are the same people saying that CS is a worthless major. CS is indeed worthless but talking shit about something you don't know about is even worse.