Sonia Poltoratski, PhD

1.6K posts

Sonia Poltoratski, PhD banner
Sonia Poltoratski, PhD

Sonia Poltoratski, PhD

@soniapolt

data for perception @zoox. previously vision neuro @stanford et al. living 😎 emoji.

Boston, MA Katılım Mart 2009
584 Takip Edilen434 Takipçiler
Cristina Cordova
Cristina Cordova@cjc·
Deciding what should exist and why is still human work.
martin_casado@martin_casado

I work with multiple companies where nearly all code is AI generated now. However, the productivity probably has only increased 20-30%. Why? I suspect because writing code is really running code. Changes are the result of a business learnings. Or an operational learnings. For mature companies, the majority of PRs are sub 10 lines codifying these learnings. AI clearly helps here (e.g. debugging, running tests, building tools) but less so. Operations and business learnings are workload and company specific. Until AI can perfectly predict what the market needs, or how a system will be used this bottleneck will exist.

English
1
1
38
8K
Daniel
Daniel@growing_daniel·
Why are CRT monitors so expensive. Why do people want them, what’s so special
Daniel tweet media
English
428
51
4.2K
477.6K
Sonia Poltoratski, PhD
Sonia Poltoratski, PhD@soniapolt·
I spent most of my twenties envious of college friends’ tech holiday parties. The wisdom of my 30s is knowing that the PPA (parahippocampal party area) event at my academic neuro conference was the more rare, special experience.
English
0
0
2
193
Sonia Poltoratski, PhD retweetledi
Tomer Ullman
Tomer Ullman@TomerUllman·
while tragic, I hope this incident ultimately leads to a new understanding of automobile psychology
Tomer Ullman tweet media
English
8
25
322
30.8K
Sonia Poltoratski, PhD
Sonia Poltoratski, PhD@soniapolt·
RT @recursus: Two things going on here: 1) Text is special. Language has evolved to convey things concisely. We choose each word carefully…
English
0
1
0
11
Sonia Poltoratski, PhD
Sonia Poltoratski, PhD@soniapolt·
baseline human performance comes up so often in my work now, but we're often constrained in what we can feasibly measure empirically. can't wait to dig in to this!
Marcel Binz@marcel_binz

Excited to announce Centaur -- the first foundation model of human cognition. Centaur can predict and simulate human behavior in any experiment expressible in natural language. You can readily download the model from @huggingface and test it yourself: huggingface.co/marcelbinz/Lla…

English
0
0
1
233
Sonia Poltoratski, PhD retweetledi
SVRHM 2022 Workshop
SVRHM 2022 Workshop@svrhm_workshop·
Congratulations to @dfinz for the NeuroAI Best Paper Award Sponsored by @NVIDIA for her paper: "Topographic DCNNs trained on a single self-supervised task capture the functional organization of cortex into visual processing streams" @eshedmargalit @cvnlab @dyamins K.Grill-Spector
SVRHM 2022 Workshop tweet media
English
1
8
63
0
Sonia Poltoratski, PhD
Sonia Poltoratski, PhD@soniapolt·
@Tweetermeyer i can correct an AV if i have good knowledge of its failure modes (like gradually learning ADAS features), or if i'm given ample time to take over (like the MB drive pilot). i will never be a reasonable safety fallback to correct a sudden veer into a cyclist in real-time.
English
0
0
2
0
Sonia Poltoratski, PhD
Sonia Poltoratski, PhD@soniapolt·
@Tweetermeyer one thing missing from this: even if humans are 100% attentive, we are not nearly as good at "the fallback role" -- correcting an AV's unpredictable and un-human-like errors -- as we are at natural driving.
English
2
1
2
0
E.W. Niedermeyer
E.W. Niedermeyer@Tweetermeyer·
There's been a lot of Autopilot/FSD discourse today, but I have to address this. I'm glad @RealDanODowd cares enough about Tesla's driving automation safety concerns to put real $$ into public education. I wish more people would! That said, his critique misses the mark for me.
@

There have been lots of requests to see the original data for this test. Further details can be found here: dawnproject.com/the-dawn-proje… We are prepared to reproduce these scientific safety tests for the media and regulators. #FSDDelendaEst

English
3
2
15
0
Sonia Poltoratski, PhD
Sonia Poltoratski, PhD@soniapolt·
my team at @perceptive_auto is hiring! we are building a new approach to AI that incorporates psychophysics and cognitive science, breaking down complex human intuition into something machines can learn and replicate at scale.
English
1
14
45
0
Sonia Poltoratski, PhD
Sonia Poltoratski, PhD@soniapolt·
I can’t wait to see everything this dataset is going to teach us about the brain. It is SO good.
@

Latest lab work in @NatureNeuro. The Natural Scenes Dataset, a deep (large-data-per-subject) 7T fMRI neuroimaging resource for vision and memory, coupled with optimized single-trial estimation methods. For people who are good with data. nature.com/articles/s4159…

English
0
0
4
0
Marc Berman
Marc Berman@AsmMarcBerman·
This week I supported SB 9 & SB 10, two bills that will help create desperately needed housing in California. Here are my thoughts 👇
Marc Berman tweet mediaMarc Berman tweet mediaMarc Berman tweet media
English
37
28
440
0
Sonia Poltoratski, PhD retweetledi
@·
I made an online tool for taking structured, searchable, and shareable notes on academic papers: paperaweek.com After 2+ years of using it myself, I'm excited to share it more broadly! Details in thread:
English
6
51
203
0

Keşfet

@cjc @growing_daniel @recursus @dfinz @nvidia @eshedmargalit @cvnlab @dyamins