I want to be at least responsible if the robots take over humanity! PhD in emergent communication (Multi-Agent Deep RL) and visions how to create understanding

Germany
Joined August 2019
Marie Ossenkopf retweeted
Surely there must be some great folks at the intersection of #audio #generativeart and #MachineLearning #art who would be into this sort of opportunity!
Paid internship opportunity this summer with @thenfb interactive art project 'The Sound of Laughter' for detecting and generating laughter. Scientific supervision by @korymath, @NeuralEnsemble, and myself. Artistic supervision by Étienne Paquette, and @thenfb. PM if interested.
2
4
1
7
Marie Ossenkopf retweeted
May our gradients never vanish. #sciencewins
5
46
3
473
Marie Ossenkopf retweeted
Writing
19
818
192
3,134
Marie Ossenkopf retweeted
Still 3 more days to submit! We also accept recent work submitted/accepted to conferences (e.g., ICLR, EMNLP, AAMAS, NeurIPS) and other workshops! 🤖📡 #NeurIPS2020
7
17
what’s better than *one* robot juggling? 🤖🦾🔴🔴🔴🦾🤖
To continue the robot juggling! Now the second arm is also moving! 🤖☺️
3
1
23
Marie Ossenkopf retweeted
Lex Fridman
@lexfridman22 Sep 2020
I make mistakes. I say stupid things. I contradict myself even when thinking deeply and rigorously. I change my mind when I learn new facts or look at ideas in a new light. I get irrationally angry. I get irrationally sad. I say things I don't mean and regret it. I'm human.
299
1,516
149
9,893
Excited to announce submissions are now open for Emergent Communication @NeurIPS 2020. This year, we tackle the problem of zero-shot communication. Submission Deadline: Oct 2nd.
6
27
Marie Ossenkopf retweeted
Another year, another workshop! We've got some cool new ideas for remote participation this year. Submit your work on language emergence, zero-shot adaptation, and more! 🤠💬🗨️🤖
7
13
Marie Ossenkopf retweeted
On point! As if relocating frequently doesnt take a toll on you. I want things in my personal life and m not willing to give them up and move just cz I want to prove that Im a good researcher. The current system makes you feel guilty for choosing this.
2
3
18
Marie Ossenkopf retweeted
Every once in awhile a paper comes out that makes you breathe a sigh of relief that you don't publish in that field... arxiv.org/pdf/2003.08505.pdf "Our results show that when hyperparameters are properly tuned via cross-validation, most methods perform similarly to one another"
31
415
63
1,606
Marie Ossenkopf retweeted
The real power move is to work in robotics, where nothing is reproducible, even by the authors 🤷‍♂️🤖
10
16
5
326
If you or anyone you know thinks they might fit the criteria listed on the flyer below, please reach out!! ** Please spread the word even if you don't meet the criteria yourself - I only need a FEW more people! Eternally grateful for every one of you. 💜💜💜
Marie Ossenkopf retweeted
Replying to @sandyasm @nlpnoah
Here's the petition for that: change.org/p/organizers-of-d…
1
2
5
Marie Ossenkopf retweeted
My bf was trying to give me a deep RL pep talk. He said, "May your half cheetah soar like an eagle and your cartpole stand for all eternity" @maxhkw
7
138