if you subscribe, you can expect all this and more, delivered straight to your timeline:
- behind-the-scenes of doing psycholinguistics research
- live regrets (a.k.a. "things I wish I'd known earlier, that are now making life hard for me")
- mild-to-warm takes on FLOSS, commercialization, and research
- lots of boosts of people asking for help, because everyone deserves to survive
- REALLY bad puns (no more than 1/day)
- existential crises (no more than 1/week)
Hi everyone! I'm a #linguistics grad student at Northwestern University, currently using #deeplearning to figure out how we learn sound patterns (#phonotactics) of new languages #phonology #phonetics #introductions
school milestones, tech (++)
2 big news this week
- I defended my prospectus! I was on probation but now I'm not going to get kicked out
- I got an internship at Duolingo for the summer! I'm down on tech a lot but this seems like the best option for "what would use my skills well, that's not academia" and I've wanted to work there since I was in undergrad so uh... we'll see how it goes.
Three days I have slogged, trying to figure out the answer to a very simple question:
How many times do you need to hear a new word accompanied by its referent before you can consistently pick out the referent, given the wordform?
It turns out that this depends on: similarity between referents, similarity between wordforms, and many other task-based factors. If I had more zeal I would propose a review article, if only to justify the slog time.
as starry-eyed as i get about language
language is a type of magic. you can *literally* transmit your private thoughts to other people
isn't that wild??????!??!
sure, sometimes it isn't 100% veridical, but the fact that we can do this type of thing *at all*, with concepts of arbitrary complexity, is... kind of amazing honestly
tips, reading, regrets
if there's one hot tip i can give to past me: write summaries of EVERYTHING you read, as separate notes in the ref manager
it was one of those things that i knew i "should be doing" but it took a lot of time and sometimes i just wanted to get through the paper in time to make it to class, okay? but even just a 1-sentence takeaway helps
i'm digging up papers that i read 2 years ago and i'm *always* relieved when there's notes
abortion rights, Alabama, US politics, donation request
The Yellowhammer Fund provides funding for anyone seeking care at one of Alabama's three abortion clinics and will help with other barriers to access (travel, lodging, etc.) as well as able.
Here's their website: https://yellowhammerfund.org/
And here's their Paypal: https://paypal.me/yellowfund
(updated w/ working link)
reinforcement learning, machine learning, artificial intelligence, evolutionary algorithms; "suicide", "cannibalism", "suffocation" (of artificial agents)
i should say, this was going around the tumblrs that I keep up with thru RSS a few days ago
reinforcement learning, machine learning, artificial intelligence, evolutionary algorithms; "suicide", "cannibalism", "suffocation" (of artificial agents)
A whole spreadsheet with examples of "specification gaming"—an AI agent hacking a reward function to find unexpected solutions: https://docs.google.com/spreadsheets/u/1/d/e/2PACX-1vRPiprOaC3HsCf5Tuum8bRfzYUiKLRqJmbOoC-32JorNdfyTiRRsR7Ea5eWtvsWzuxo8bjOxCG84dAg/pubhtml
fave: In a game meant to simulate the evolution of creatures, the programmer had to remove "a survival strategy where creatures could gain energy by suffocating themselves" (Schumacher, 2018)
ACL conference ++
in honor of this, i've made my repo public: https://github.com/nimirea/phonRNN
apparently my fonts were just, like, mis-specified this whole time and i had no idea because my default fonts were pretty
now with less Times New Roman!
git, social science, blog post
did something i've been threatening to do for Years and started a """professional blog"""
the first post is about git and social science
it has stills from Repo Man
https://www.nimirea.com/blog/2019/05/10/git-for-social-scientists/
used to be really impressed at really long lit reviews that cite like 50 papers. like how can you write multiple articles a *year* like that??
turns out these people have been citing the same 40 papers in every article, and just keeping up with the latest research and incorporating that as they go along
still impressed, but it doesn't seem impossible now :)
Your reminder that a fellow mastopal needs help funding her Masters'
Just £7 to hit £500!!
https://www.gofundme.com/help-to-fund-the-last-6-months-of-my-masters
RT @MIT_CSAIL@twitter.com
Left: MIT computer scientist Katie Bouman w/stacks of hard drives of black hole image data.
Right: MIT computer scientist Margaret Hamilton w/the code she wrote that helped put a man on the moon.
(image credit @floragraham@twitter.com)
Los Angeles, politics, surveillance
ok so i know there are at least a couple of people following me who live in LA. check this out:
basically, LADOT wants to force dockless mobility operators (scooter-/bike-share companies, like Lime and Bird) to send them realtime location data w/o being clear about how it'll use the data, how long it'll keep it, and the conditions on which it shares data with third parties (which it's already done).
animal shelter, asking for donations
just gonna go ahead and plug Evanston Animal Shelter's donation page real quick: http://evanstonanimalshelter.net/donate-funds/
the shelter is especially in need after recent events (FB link, warning: sad story about dog) https://m.facebook.com/549032201/posts/10155783832252202/
Linguistics grad student at Northwestern University, studying language acquisition, bilingualism, and cognition. Many preoccupations. This is one facet of many.