machine learning: this place is not a place of honor

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > YOSPOS > machine learning: this place is not a place of honor

fart simpson: Jul 2, 2005; DEATH TO AMERICA

Sagebrush posted:

To me the most damning example is those papers with adversarial techniques where they apply a really subtle filter to the image and it tricks the algorithm. Like you start with an image of a turtle, and the machine recognizes it as a turtle, and then you apply a tiny convolution that affects 10% of the pixels in an almost undetectable way, and the algorithm is now certain that it's a picture of a gun. But it still looks like a turtle to you.

So the takeaway is that whatever the algorithm is triggering on, it's absolutely not what humans think of as turtle-like features. It doesn't see things the way that humans do, doesn't have the same "mental" model of the world. That means you can't make any assumptions about its behavior that are based on how humans would treat a given situation. yet that's exactly what people do when they start pressing these systems into service in things like self-driving cars.

wow, if this is true, you should contact the media. could be a big story

# ¿ May 30, 2019 07:45

Adbot: ADBOT LOVES YOU

# ¿ May 11, 2024 11:59

fart simpson: Jul 2, 2005; DEATH TO AMERICA

rotor posted:

For real though anyone doing research in ML or machine vision should stop. The downsides of these technologies significantly outweigh the benefits.

# ¿ Jun 4, 2019 01:39

fart simpson: Jul 2, 2005; DEATH TO AMERICA

# ¿ Sep 21, 2019 05:01

fart simpson: Jul 2, 2005; DEATH TO AMERICA

Salt Fish posted:

It's an honest mistake, you see the car is programmed to hit cyclists at 45mph and it got confused.

well someone in that thread said it had a 1 second delay before braking to avoid false alarms. so every reclassification triggered an additional 1 second delay

# ¿ Nov 8, 2019 09:27

fart simpson: Jul 2, 2005; DEATH TO AMERICA

Sagebrush posted:

only in driving ability, which doesn't define the person.

my grandma is a wonderful person but god drat for the last few years before she gave up driving it was absolutely terrifying to be in the car with her.

i know another woman who's very smart and competent in day to day life but her driving has made multiple people i know carsick to the point of vomiting.

some people just aren't able to drive and that's fine. only in america is it treated as some sort of human right

in america it�s a human necessity

# ¿ Nov 8, 2019 19:09

fart simpson: Jul 2, 2005; DEATH TO AMERICA

echinopsis posted:

the difference between skilled/experienced drinkers and amateurs is the ability to focus and concentrate. amateurs get distracted their hosed mind where pros can concentrate hard enough to survive

lol

# ¿ Nov 11, 2019 13:17

fart simpson: Jul 2, 2005; DEATH TO AMERICA

Notorious b.s.d. posted:

it didn't use to be

we can turn back the clock on this one

it's not like operating bus routes is some forgotten technology, impossible to replicate in the modern age

it very well may be impossible to replicate in the modern american political environment

# ¿ Nov 11, 2019 13:18

fart simpson: Jul 2, 2005; DEATH TO AMERICA

Cybernetic Vermin posted:

a lot of ml research is dumb and bad (almost as dumb and bad as "tech visionaries" ideas of what ml can do), amounting to a phd student twiddling various parameters until a model seems to learn something, with no deeper analysis or insight. however, it is not right to view ml as having been bad for computer vision and natural language processing, not least ml is pretty simple, so it is not a hard tool to apply.

i've historically been in nlp myself, and while its been a weird decade ever since statistical n-gram methods broke the backs of the chomsky'ites things are looking better and better now. it has gotten clear which ml bits are indispensable and possible to analyse (e.g. word embeddings) and the field is just getting a lot more high-level thanks to ml bits handling a lot of the nitty-gritty (e.g. making it a lot easier to make stuff robust against grammatical mistakes). it has chilled interest in formal grammars and automata solutions a lot, but there was so much theoretical navel-gazing there that i don't think that's bad (and that was my primary research area).

i am also currently pretty excited about a new research program from the research group i primarily affiliate with, where they are embracing the bias-soaking nature of ml to study gender bias in written text. that is, as a very first step, looking at the word embeddings for ostensibly non-gendered words in a given publication and see how orthogonal those feature vectors are from gendered vectors. a ton of tricky issues there (e.g. it matters a lot how the dimensionality reduction, i.e. the ml, works), but a bright new phd student (affiliated both with us at cs and the dept. of gender studies) is working on it, and i think it'll be extremely interesting research no matter the exact outcome.

well, that's a long post. tl,dr: ml *in* research often good.

is nlp still basically only done in english

# ¿ Jan 3, 2020 17:51

fart simpson: Jul 2, 2005; DEATH TO AMERICA

i was looking around for some software to segment chinese text into words and it seems anything with more than 90% accuracy is like cutting edge university research algorithms

the most popular one people in china actually use is so bad that i, a non native speaker, can find errors in basically every sentence i throw it at

fart simpson fucked around with this message at 02:26 on Jan 4, 2020

# ¿ Jan 4, 2020 02:18

fart simpson: Jul 2, 2005; DEATH TO AMERICA

idk maybe? i dont really know anything about linguistics but i do know you can give a chinese speaker a sentence and they can easily split it into component words. i mean it is more complicated in chinese because words can have sorta layered meanings in a way so maybe that means it's an artificial idea? but like, what seems to be the most popular tool that people in china use for this is a python library called jieba. i downloaded jieba and the very first sentence i threw into it was the first sentence from the chinese wikipedia article for "flower":

花是被子植物的繁殖器官
(flowers are the reproductive organs of angiosperms)

if you asked any chinese reader they'd come up with:
花 / 是 / 被子植物 / 的 / 繁殖 / 器官
flowers / are / angiosperm / (posessive marker) / reproductive / organs

jieba segmented this as the obviously nonsensical:
花是 / 被子植物 / 的 / 繁殖 / 器官
flowersare / angiosperm / (possessive marker) / reproductive / organs

actually google translate does word segmentation too and also fails even worse than jieba on this sentence although it gets the overall meaning correct so i guess i see your point about the statistical methods thing

# ¿ Jan 4, 2020 12:47

fart simpson: Jul 2, 2005; DEATH TO AMERICA

yeah i was surprised too. i think there's probably some settings i can adjust because i found a javascript reimplementation of jieba that gets this sentence correct. but yeah i'm playing around with a dataset i found of 300k news articles written in chinese and just this usage of "是" as a verb makes up 1% of the entire body of text of the dataset. it's probably either the 1st or 2nd most commonly used verb in chinese.

google's segmenter got that word correct but totally butchered the segmentation of angiosperms into 3 separate words which would translate as like, "blanket seed plants" or something which i guess is kinda what angiosperms are anyway? especially because the word 被 can be a noun meaning blanket or a verb meaning to cover

# ¿ Jan 4, 2020 13:52

fart simpson: Jul 2, 2005; DEATH TO AMERICA

i went to music school and our weeder courses were ear training and music history

# ¿ Mar 27, 2020 16:25

fart simpson: Jul 2, 2005; DEATH TO AMERICA

Jabor posted:

That's because you've never talked to anyone who honestly works with stats.

Yeah, there are a whole bunch of liars who set out to use the formalisms of stats to justify whatever agenda they're trying to push, and also a bunch of idiots that throw stats machinery at problems and blindly pick things with a strong correlation.

A good statistician doesn't do either of those things though.

lol

# ¿ Jun 17, 2020 06:41