How much interest do you guys have in data science and machine learning? The few times I pop in, I never see anything in the catalog. Maybe it's a little pleb-tier for this board? I think it'd be cool to have a persistent thread about it though.
We could even have interesting activities, like friendly-stalking some /ic/ general and generating art in someone else's style, or other predictive stuff on some board. The Veeky Forums api is pretty good, I think that could be amusing for all and more fun than playing with cancer rate datasets or whatever. Also of course discussion about ML techniques, libraries, projects, etc.
Just a thought, also what do you think about machine learning?
data science is one of those fields that should be intensely interesting but actually just ends up being filled with bottom feeding CS troglodytes who think they understand statistics because they got a B- in their "statistics for computer science majors" course
Parker Turner
In September, I'm expecting to start a PhD in machine learning and pattern recognition for bioinformatics and health informatics, so I'm quite interested.
I've studied data science, and while it could be an interesting field to work in, it still has a way to go as it is filled with plebs who are just looking to make money, yes. The field really needs to become better defined; less about playing with a set list of algorithms to see which ones work, less about making pretty graphs for business people, and less about business bullshit. There should be data science teams; rather than expecting to find all the 'data science skills' in one person, and expecting them to find the time to do set up a data warehouse, play with algorithms, make visualisations, and give presentations to a room full of business majors (who barely understand the equation of a straight line) and try to convince them that data is better than their intuition. The trick is either to get the interesting roles in the data science team, or get brought on as a consultant. That would be the most interesting part.
Jace Nguyen
I somewhat agree with you, but I don't think it has to be that way. And even bottom feeding CS troglodytes might enjoy learning more, and talking about techniques and whatever. I appreciate your spirited response!
Oh that's fantastic. My background is Math BS, that is slowly rotting away from disuse. And I don't particularly love programming, but I've come to accept that I need to. And I like programming in R actually. I was surprised that the techniques in ML were so simple, mathematically, but still yielded pretty interesting results.
Totally into what you're saying. I'm assuming you know of kaggle. Maybe we could gather interest in a Veeky Forums kaggle team, with proceeds donated somewhere to keep life simple (implying we'd win).
>try to convince them that data is better than their intuition This is always the kicker. Being able to display with absolute certainty, superiority over human intuition, then becoming an indisputably useful tool.
Dylan Perry
Not too much interest though? Not enough to sustain a general, right?
bumping for your thoughts
Camden Edwards
Where can i get info on accessing the Veeky Forums api?
Gavin Wright
by googling "Veeky Forums api" and clicking the first link (obligatory sass, immediately regretful, sorry friend)
I'm physics background myself, but I'm not sure the maths you learn is completely unused; I think the way of thinking it trains you with stays with you forever. It's good to hear you're interested in R; you may find something fun in functional programming. Machine learning is an interesting field for the reason that the difficulting is in implementing the same techniques in new ways; you're often applying them to new problems that haven't yet been tackled, or adding an SVM to a CNN, or something quite interesting I came across recently was visual question answering, where you use a CNN in segmented images and a bag-of-words for associated descriptions, and you are able to feed images into this classifier and ask it questions like "where is the cat?", and get natural language responses like "on top of the laptop" (arxiv.org/pdf/1512.02167.pdf). I think a Veeky Forums kaggle team would be a bad idea, honestly... especially if money became involved!
Ethan Powell
I agree that math stays with you in a certain way. For example, now I'm able to teach myself new maths, or easily relearn the subjects I used to know. In a way, I think that's what my degree gave me. I love R, and functional programming. I liked haskell but ditched it after one too many difficulties with cabal/stack. R can be a little idiosyncratic, and maybe it's just that I've been getting used to it (and the greatness of CRAN and rstudio), but I have fun programming in it.
That classifier sounds awesome. I'm really excited to get more into actually building stuff, so far having only done some handwriting recognition and similar things, guided heavily by the coursera. That looks like a pretty great paper, I'll add it to my pile of ML links that I'll probably never read. Ahhh, too much to learn in a lifetime.
I think there's not enough interested traffic here in Veeky Forums to support a persistent machine learning general, so it's being discussed on /g/ which may be more appropriate, though I'm worried it's a little fast there. Well, we'll see how it goes! Currently just hijacked a joke thread about NNs to try and get an indication of interest. In a short time, a number of people seemed interested, which leads me to believe there would be a fair amount of interested people, and newcomers, to support it there.
Thank you, sass deserved, now I have another googleable question- are you aware of any research that has used this api? I'd like to see what has been done before i dive in.
David Walker
Well, those are good resources, but imageboards are my favorite way to internet by far. And I like the thought of a cozy Veeky Forums ML community. Plus, I've already worked a little with the Veeky Forums API and a community doing analytics on itself (and other Veeky Forums communities) just sounds fun as hell. Also people are less afraid of "looking stupid" whatever that means, when they're anonymous (yes, there's some anonymity in user names, but not really in practice.) So, I'll give my best shot at the /mlds/ general on /g/.
Please stop by if it actually takes off!
I'm not aware of any, maybe you should ask in /mlds/ on /g/ when it gets started, or for now :)))))) I would love running some stats, data analytics on Veeky Forums. Especially because I really don't know how to do that sort of thing very well, but I'd like to. I was working on something for the near-mythically-shitposted amateur game development general on Veeky Forums, to classify posts as "shitposts", "progressposts", or "helpfulposts." I got about as far as setting up the scaffolding to grab and tag specific posts when I gave up. Some of the API is a little weird, but mostly it's clear.
Let me know if you find anything (especially if it's in R), I'd be interested too.
I'm currently working on an AI chatbot that makes use of machine learning, I have an NN written in lua that I use to train the bot. It's fairly intelligent but I've hit a wall due to my poor hardware.
Alexander Rivera
>CS major here, focused in nlp
>I somewhat agree with you, but I don't think it has to be that way. And even bottom feeding CS troglodytes might enjoy learning more, and talking about techniques and whatever. I appreciate your spirited response!
I love when other majors believe that they can usurp the profession of ""data scientist"", profession that was tailored by computer science. It's funny when other majors claim like they can be """huuurrr data scientists""", give them a dataset and they just can't figure out how they are going to munge all the data or how to do a cross validation.
>I was surprised that the techniques in ML were so simple
Sure faggot, there are several open problems... why dont you bring a solution and get rich?.
>It's good to hear you're interested in R >I'm physics background myself
R is a shit programming language, typically used by people who can not handle state of the art tools like python or in some cases julia. Try to parse or POS-tag a corpus with R faggot, it will take a long.
Also, nice CS book :) [pic related]
Alexander Moore
Holy dang, /g/ is actually taking to this.
It looks like it could hold.
If you're into ML, data science, AI, we'd love to have you. (also don't even lie, the birth of a new general is kind of exciting)
Or hopefully, search catalog for /mlds/ in the future.
Noah Brown
>R is a shit programming language, typically used by people who can not handle state of the art tools like python or in some cases julia. Try to parse or POS-tag a corpus with R faggot, it will take a long. Except you though, your needlessly aggressive attitude is boring, you're not invited.
I've worked with C, C++, java, haskell, javascript, fortran, R, python, ruby, probably some others. Just because I don't like programming doesn't mean I haven't done a lot of it. And I think the POV here that I've quoted is boring. I won't hold you to it though, we're all anonymous here. I just think you should reconsider your "strong opinions" :)
Blake Wright
whats the story on the picture?
Jason Harris
Does this course look worthwhile to any of you? My background is a Mathematics Bsc. It included some statistics and probability, but I mostly went with pure modules.
That thread died with 94 posts and 14 unique posters
That's fucking pathetic, stop trying to force this crap if no one is interested
Henry Richardson
I have no interest in forcing stuff nobody is interested in, that's why I gave it a shot, but I'm not making a new one.
Nolan Brown
But also don't act like it was such a bad idea. There's a bit of interest.
Juan Gutierrez
>I would love running some stats, data analytics on Veeky Forums. Never fear, the nsa is already doing this.
Ian Scott
Nah, dude. R simply fucking sucks.
Jackson White
Cool discussion bro!
Bentley Allen
Thanks for thew thumbs up! You made my day!
t. the only person on Veeky Forums with a top ten kaggle finish
Jacob Morales
Chill out, nerds.
Landon Brooks
dont call me a nerd I have black belt
Jack Murphy
Just say you're too lazy to argue and move on, trying to appeal to authority just makes you seem pathetic
Camden Price
w-whoa sorry for what I said
Justin Richardson
I drink like 3 red bulls a day dude try me
Angel Gomez
I'm too lazy to argue. That doesn't change the fact that R sucks big hairy donkey balls.
Aiden Moore
>muh AbstractObjectPatternContainerFactory
Brandon Jones
The reason I would not be interested in a Veeky Forums kaggle team with you is primarily because you obviously don't know shit and I don't feel like babysitting your dumb ass. What do you bring to the table? You're one of the bottom feeding troglodytes of which you are so dismissive. Fuck off, poseur.
Jack Johnson
Is that what has you pouty? I was using that term because the other user did, actually kind of mocking him for using it, but I suppose that didn't come through in text.
Love this enthusiasm though!
Gavin Reed
I also use assembly for data analysis, fellow scibro.
Oliver Scott
What is the mathematics behind theoretical machine learning? I heard that theoretical machine learning is just math.
Carter Brooks
>I heard that theoretical machine learning is just math. You heard wrong it's actually just poetry
Jackson Hughes
Most of the entertaining math in Ml involves setting up and solving optimization problems. Think linear algebra, numerical methods, multivariate calculus , probability theory, mathematical statistics.
Jason Powell
V A P N I K A P N I K
Jackson Rogers
basically some linear algebra and calculus.
You could probably involve some higher level things like control theory but mostly it's just fancy regression.
Because those who have interest in machine learning are on /g/
Asher Brown
You'd think so but my attempt at a machine learning/data science general died very quickly, and only stayed alive with me samefagging like a samefag. Then an /ai/ general started, and might come back, but it died too.
I feel like it's in a sad spot between Veeky Forums finding it too pleb, and /g/ being too fast and concerned with other things.
Ethan Rivera
Veeky Forums knows how to spot a meme field
/g/ is just plain too stupid
Samuel Bell
Even McCarthy spotted an actual communist every so often.
