Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us $3,400 per month for bandwidth bills alone, and since we don't believe in shoving popup ads to our registered users, we try to make the money back through forum registrations.
  • Post
  • Reply
animist
Aug 28, 2018

yeet





machine learning is when you have a bunch of data and you want to tune your system based on that data. it's related to data analytics, which is when you just want to figure things out about the data, and statistics, which is the boring, stuffy version of data analytics, for people who care about "math" and "proofs" and "hundreds of years of scientific thought". machine learning is also related to AI, which is 20th-century parlance for "doing things with computers".

in the past decade there's been an explosion of popularity of machine learning, mostly due to cheap compute + lots of large datasets floating around, now that everyone posts everything they're doing on the internet all the time.

machine learning is extremely popular with tech monoliths like google, amazon, netflix, etc. because it helps them squeeze money out of aloof millenials. if you see people talking about the "google algorithm", they're actually talking about frankensteins-monster ML systems built out of ad-hoc mixes of hundreds of different ML models, for the sole purpose of pushing pixels and maximizing adbucks.

i am doing machine-learning related stuff for the forseeable future and am posting this in the hope that some of you want to commiserate. feel free to post whatever ML-related bullshit in this thread. i can effortpost about deep learning interpretability research if people are interested, that's my area.

FAQ:

Q: how do i machine learning?
acquire a large amount of illegally harvested user data. pour it into a poorly-tested python codebase. trust whatever the system spits out. publish at least 3 papers

Q: which machine learning should I use?
  • i want to predict bits of my illegally harvested user data based on other bits of my illegally harvested user data
    -> use xgboost
  • i want to group my illegally harvested user data into a bunch of categories based on criteria i don't understand
    -> use k-means or DBSCAN
  • i want to arrange my illegally harvested user data into a bunch of vaguely suggestive two-dimensional blobs
    -> use TSNE
  • i want to validate my data against a model and attribute fractions of variance to different factors of the model
    -> "validate" a "model"? what is this, some kinda fancy-pants science-party? go back to academia, statistics nerd. this is the wild west

Q: how do i run that machine learning on my learning machine?
pip install numpy pandas matplotlib jupyterlab scipy scikit-learn xgboost

Q: what is deep learning?
deep learning is when you use commodity GPUs and gradient descent to train enormous nonlinear functions. deep learning is only useful for speech synthesis and niche pornography.

Q: is machine learning going to cause the singularity?
no. machine learning exists exclusively to allow hiring managers and advertising executives be racist slightly more efficiently

Q: how do i acquire machine learning figgies?
get a phd. sorry, i don't make the rules

Q: what is the 'kernel trick'?
a sex act performed by talented professionals in a particularly exclusive gentleman's club on the outskirts of Reno

animist fucked around with this message at 21:22 on May 25, 2019

Adbot
ADBOT LOVES YOU

Skim Milk
Nov 19, 2006



we can teach a machine to learn, but can we teach a machine to love?

thebigcow
Jan 3, 2001

Bully!

"I got that reference" but it's a cybernetic Captain America from some Marvel one off.

DONT THREAD ON ME
Oct 1, 2002

by Nyc_Tattoo


Floss Finder

how do we do anti-machine learning things like how do we create more bogus user data that will gently caress with machine learning poo poo

animist
Aug 28, 2018

yeet



DONT THREAD ON ME posted:

how do we do anti-machine learning things like how do we create more bogus user data that will gently caress with machine learning poo poo

i always lie on the youtube quizzes that show up sometimes

there's also adversarial examples which are basically optical illusions for attacking deep learning systems:



those are pretty crazy, like you can wear pixelly facepaint at a protest to trick facial recognition systems

TheFluff
Dec 13, 2006

FRIENDS, LISTEN TO ME
I AM A SEAGULL
OF WEALTH AND TASTE


Slippery Tilde

obligatory ceglowski quote incoming

quote:

[T]reating the world as software promotes fantasies of control. And the best kind of control is control without responsibility. Our unique position as authors of software used by millions gives us power, but we don't accept that this should make us accountable. We're programmers—who else is going to write the software that runs the world? To put it plainly, we are surprised that people seem to get mad at us for trying to help.

Fortunately we are smart people and have found a way out of this predicament. Instead of relying on algorithms, which we can be accused of manipulating for our benefit, we have turned to machine learning, an ingenious way of disclaiming responsibility for anything. Machine learning is like money laundering for bias. It's a clean, mathematical apparatus that gives the status quo the aura of logical inevitability. The numbers don't lie.

Shaggar
Apr 26, 2006


Nap Ghost

animist posted:

i always lie on the youtube quizzes that show up sometimes

there's also adversarial examples which are basically optical illusions for attacking deep learning systems:



those are pretty crazy, like you can wear pixelly facepaint at a protest to trick facial recognition systems

the government has probably moved on to gait matching anyways

jimmyjams
Jan 10, 2001


King Kong of Megadongs
Gobblin' them mega schlongs
Makin' sure they mega long
Stroke' 'em if they mega strong


isnt that also hilariously defeatable by simply putting a rock in your shoe

power botton
Nov 2, 2011



we do ~*UBA*~ software and after years of management crying we don't have machine learning and can't sell anything without it, we finally released something and barely any of our customers give a poo poo aside from a checkmark on an RFP.

animist
Aug 28, 2018

yeet



power botton posted:

we do ~*UBA*~ software and after years of management crying we don't have machine learning and can't sell anything without it, we finally released something and barely any of our customers give a poo poo aside from a checkmark on an RFP.

it's well known that customers care deeply about the implementation details of the software they use

Fiedler
Jun 29, 2002

I, for one, welcome our new mouse overlords.

if your job is to purchase enterprise software from ISVs, you learn very loving quickly to care about the implementation details

rotor
Jun 11, 2001

Official Carrier
of the Neil Bush Torch

 
 
 
 
teh butts


Fiedler posted:

if your job is to purchase enterprise software from ISVs, you learn very loving quickly to care about the implementation details

i cannot reconcile this statement with a lifetime of working with and on enterprise software.

eschaton
Mar 7, 2007
Ask me how it all ends!

Oven Wrangler

something needs to provide executive function on top of all the pattern marchers people have been building

symbolic AI is coming back for sure

2019 year of cyc

Fiedler
Jun 29, 2002

I, for one, welcome our new mouse overlords.

executing the pattern matchers sounds like a good plan to me

power botton
Nov 2, 2011



animist posted:

it's well known that customers care deeply about the implementation details of the software they use

for certain things we advertise they certainly do and really put us through the paces (Mainly pass the hash detection). for the ML bullshit they're like oh great, I don't even have to see it generate anything during the eval, at least I can get my boss off my back saying you guys to do it.

power botton
Nov 2, 2011



we use HDBSCAN btw

Captain Foo
May 11, 2004

we vibin'
we slidin'
we breathin'
we dyin'


Clever Betty

this is the thread where i post ML is money laundering but for bias and then don't attribute it because i don't remember where it came from, right

DONT THREAD ON ME
Oct 1, 2002

by Nyc_Tattoo


Floss Finder

i'm going to start a new branch of computing called woke algorithmics

DONT THREAD ON ME
Oct 1, 2002

by Nyc_Tattoo


Floss Finder

i think if we teach the computers about intersectionality the bias problem will just go away

huhwhat
Apr 22, 2010


humans are a bad example to learn from imo, we should focus on machine teaching and learn the emotionless ways of 1s and 0s from our robotic superiors

DONT THREAD ON ME
Oct 1, 2002

by Nyc_Tattoo


Floss Finder

https://twitter.com/yiqinfu/status/1133215940936650754

i call this the 'i'm single-arity'

huhwhat
Apr 22, 2010


wonder if we'll see a surge in amateur porn uploads if pornhub automatically deepfakes the faces to some long dead pornstars' before publishing the vids

flakeloaf
Feb 26, 2003

Still better than android clock




no part of this is good

DONT THREAD ON ME
Oct 1, 2002

by Nyc_Tattoo


Floss Finder

huhwhat posted:

wonder if we'll see a surge in amateur porn uploads if pornhub automatically deepfakes the faces to some long dead pornstars' before publishing the vids

there should be one where you can upload pictures of your own face and then you can jack off to porn where the male pornstars faces have bene replaced with your own

DONT THREAD ON ME fucked around with this message at 17:04 on May 28, 2019

power botton
Nov 2, 2011



except it should be for penises so i can pretend i can satisfy a woman with my tiny little pecker

DONT THREAD ON ME
Oct 1, 2002

by Nyc_Tattoo


Floss Finder

power botton posted:

except it should be for penises so i can pretend i can satisfy a woman with my tiny little pecker

ithought about that too, it's going to be a stretch goal

Bloody
Mar 3, 2013



machine learning is matrices with racism inside

Bloody
Mar 3, 2013



http://oneweirdkerneltrick.com/

Captain Foo
May 11, 2004

we vibin'
we slidin'
we breathin'
we dyin'


Clever Betty

Bloody posted:

machine learning is matrices with racism inside

poty
Jun 21, 2008

虹はどこで終わるのですか? あなたの魂の中で、または地平線で?


i didnt read the op because its long and im sleepy but i will soon

Silver Alicorn
Mar 30, 2008

dude put that thing away


I dident read it and I never will

suffix
Jul 27, 2013

Wheeee!


DONT THREAD ON ME posted:

there should be one where you can upload pictures of your own face and then you can jack off to porn where the male pornstars faces have bene replaced with your own

plot twits: you only watch male j/o videos anyway

DONT THREAD ON ME
Oct 1, 2002

by Nyc_Tattoo


Floss Finder

Bloody posted:

machine learning is matrices with racism inside

suffix
Jul 27, 2013

Wheeee!


i like the funny markov rnn bots it's a shame that they'll trigger the butlerian jihad soon when the net is 98% troll bots

Helianthus Annuus
Feb 21, 2006

can i touch your hand


Grimey Drawer

DONT THREAD ON ME posted:

ithought about that too, it's going to be a stretch goal

laughing about a penis stretch goal. thx op

Sagebrush
Feb 26, 2012





Gravy Boat 2k

DONT THREAD ON ME posted:

there should be one where you can upload pictures of your own face and then you can jack off to porn where the male pornstars faces have bene replaced with your own

if you do this you'll be a trillionaire

Asleep Style
Oct 20, 2010



Helianthus Annuus posted:

laughing about a penis stretch goal. thx op

Every morning I wake up and open palm hang a 50 pound weight off the end of my dongus

DONT THREAD ON ME
Oct 1, 2002

by Nyc_Tattoo


Floss Finder

Sagebrush posted:

if you do this you'll be a trillionaire

i'd only do it if there were a way to prevent people from using it to deep fake other people.

maybe you could engineer some sort of placard that you could mail them and then they could hold it up in their photos to verify their id or something.

e: yeah i dont think you could prevent abuse.

DONT THREAD ON ME fucked around with this message at 03:39 on May 29, 2019

flakeloaf
Feb 26, 2003

Still better than android clock



Asleep Style posted:

Every morning I wake up and open palm hang a 50 pound weight off the end of my dongus

Sagebrush posted:

if you do this you'll be a trillionaire

Adbot
ADBOT LOVES YOU

pointsofdata
Apr 25, 2011



we replaced a vendors "ml" process with an afternoon writing regexes and got better result. seems like a rip off op, I wouldn't buy again.

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply