Kim Stachenfeld @neuro_kim Twitter profile

Last Seen Profiles

@LaylaScerr46684

@pengen_stw

@jandakembangstw

@johanbrand

@Kellyfish636

@ugo_ew

@THIAZINSINCERO1

@robbielab

@Fifu

@JoeT_Radio

@Tin0din0

@M3tering

@mochisuhs

@muhammadsadamh

@GFK_Eagle

@pengen_stw

@Nhkstchen3

@Roterteufel89

@lliga_d

@LBaghyshny

@Foxy_on_the_run

@AminYakoub52309

@Janniklas1998

@Hason818

@MOVIRSV

@myluteal

@jandakembangstw

@feedback_King

@Tinajam14043176

@KathyMo55649064

@jandakembangstw

@chucky_beatz

@SkankHunt69690

@mongerr27

@BeeGatman

@jandakembangstw

Kim Stachenfeld

@neuro_kim

2 years

We have a paper in ICLR! The title is “Learned Coarse Models for Efficient Turbulence Simulation.” We wanted to see if we could train general-purpose ML models to predict turbulent dynamics accurately at low spatial and temporal resolutions (1/n).

8

93

443

Kim Stachenfeld

@neuro_kim

5 years

Slide deck from recent @CompCogNeuro Tutorial with @behrenstimb on "Representing States and Spaces." 4.5 hours, ~200 slides and we covered almost 2/3 of them! A public-speaking Everest :) 1/3

7

75

267

Kim Stachenfeld

@neuro_kim

2 years

DeepMind internships are opening this Friday Sep 16 and will remain open for ~2 weeks! Feel free to get in contact with me if you have questions.

Google DeepMind

Artificial intelligence could be one of humanity’s most useful inventions. We research and build safe artificial intelligence systems. We're committed to solving intelligence, to advance science...

deepmind.google

9

70

255

Kim Stachenfeld

@neuro_kim

3 years

Excited this work by Dan McNamee, me, Matt Botvinick, and @gershbrain is out! Link without paywall:

Google DeepMind

@GoogleDeepMind

3 years

Cognitive processes like exploration, memory consolidation & planning use sequences from a mental model. New work in @NatureNeuro shows how hippocampal-entorhinal neurons can be easily modulated to generate sequences optimised for different cognitive goals

2

158

645

3

40

195

Kim Stachenfeld

@neuro_kim

6 months

Applications open for student researchers (formerly interns)! There will likely be projects with me+colleagues (including @kevinjmiller10 !) on comp neuro, and ML directions on memory+retrieval. Locations in NYC/London. Reach out if any questions!

Careers

At the heart of Google DeepMind’s mission is our commitment to act as responsible pioneers in the field of AI, in service of society’s needs and expectations.

deepmind.google

Kevin Miller

@kevinjmiller10

6 months

Applications are open for the @GoogleDeepMind Student Researcher Program! There will likely be projects available to work with me and with others in computational neuroscience. If interested, please feel free to get in touch! Learn more and apply here:

6

100

455

7

38

184

Kim Stachenfeld

@neuro_kim

4 years

i've been looking for reviews+textbooks in neuroscience that emphasize historical context/the evolution of ideas in a neuro fields. do people have any favorites to recommend?

22

21

157

Kim Stachenfeld

@neuro_kim

1 year

Tremendously excited to be giving an RL tutorial at #cosyne2023 ! Materials for the tutorial can be found here. slides coding colab (File > Make a copy; Runtime > Change runtime type > GPU)

CosyneTutorial_RL_2023_Stachenfeld.ipynb

Colab notebook

drive.google.com

5

30

147

Kim Stachenfeld

@neuro_kim

7 months

The (brand new!) AI Institute ARNI (ARtificial + Natural Intelligence) is hiring postdocs to work at the intersection of AI+neuro, at Columbia U. Please share! Info: Apply: (reviewed on rolling basis)

0

47

141

Kim Stachenfeld

@neuro_kim

2 years

We had a paper accepted to ICLR 2022! I'm very excited about this, and my next post will be about that. But first, I feel I need to post a tribute to the recently deceased musical artist Meatloaf. (1/n)

2

5

121

Kim Stachenfeld

@neuro_kim

7 years

DeepMind's blog post about our paper on predictive representations in hippocampus!

Google DeepMind

@GoogleDeepMind

7 years

RL meets neuroscience in our new @NatureNeuro paper: the hippocampus as a predictive map

8

280

531

0

20

96

Kim Stachenfeld

@neuro_kim

2 years

Very excited to share this new work with @ZePoLiTaT and @KelseyRAllen . Motivation: we want AI that can invent new things! In this work, we use fancy new learned simulators (made of Graph Neural Networks) to optimize designs in varied settings with complex dynamics.

Kelsey Allen

@KelseyRAllen

2 years

Could machines design new tools for us? With my co-first authors @neuro_kim and @ZePoLiTaT , we are excited to announce "Physical Design using Differentiable Learned Simulators" where we solve complex, physical design problems by differentiating through GNN-based simulators. 1/9

9

200

1K

1

12

91

Kim Stachenfeld

@neuro_kim

5 years

Utterly honored to be included on this list with all these super cool people. Thank you @techreview #35InnovatorsUnder35

MIT Technology Review

@techreview

5 years

Announcing the new class of #35InnovatorsUnder35 . Past honorees include Mark Zuckerberg, Sergey Brin, Larry Page, and Jonny Ive. This crew should be just as successful. Here's your chance to meet them first.

10

124

278

2

5

92

Kim Stachenfeld

@neuro_kim

4 years

trying to get work done today like

2

0

82

Kim Stachenfeld

@neuro_kim

8 months

This was a fantastically fun conversation with (fellow neuroscientist!!) @JonKrohnLearns about neuroscience, memory, and the possibilities for ruminative AI that is as anxious and neurotic as us! Should be available end of October on the SuperDataScience podcast.

Jon Krohn 🇺🇦

@JonKrohnLearns

9 months

Next week, I'm interviewing Dr. @neuro_kim — Research Scientist at Google DeepMind and Affiliate Professor of Theoretical Neuroscience at Columbia University — for a podcast episode. Got questions for her? In addition to the above, Kim: • Is an exceptional and fun explainer

1

2

22

0

3

64

Kim Stachenfeld

@neuro_kim

6 years

I got my neuroscience blackbelt!

Ida Momennejad

@criticalneuro

6 years

Everyone, @neuro_kim is a doc now & she has this hippocampus poster signed by all PNI to show for it! She's that friend I share science thoughts, conference rooms, & life stories with - for 5ish years!! Here's to 50 more years of science & friendship! #successorrepresentation

2

58

5

1

58

Kim Stachenfeld

@neuro_kim

6 years

Organizing a workshop on "Model-Based Cognition: Hierarchical Reasoning and Sequential Planning" for #cosyne2018 with @roozbeh_kiani @basvanopheusden @kevinjmiller1 . We have gathered some great speakers and you should come hang out with them! See more:

3

13

54

Kim Stachenfeld

@neuro_kim

5 years

Lovely article about the state of research on neural representations of space and not space. Especially lovely is that I'm in it!

Quanta Magazine

@QuantaMagazine

5 years

The evidence is growing that the brain’s grid cells, used to identify our position in physical space, may also be keys to organizing and navigating more abstract concepts, like time and memory.

3

45

115

1

11

51

Kim Stachenfeld

@neuro_kim

8 months

I was on a PODCAST, how cool is that? Hear me wax lyrical on the neuroscience + AI of learning and prediction. Find fresh takes, like maybe LLMs are kind of a big deal (...but there are still some open problems for the rest of us). Thanks @superdatasci for the interview!

Jon Krohn 🇺🇦

@JonKrohnLearns

8 months

Today's episode with is one of my favorite conversations ever. In it, the hilarious and fascinating Dr. @neuro_kim (of both DeepMind and Columbia) blows my mind by detailing relationships between human neuroscience and A.I. Watch here: More on Kim: •

0

3

13

2

5

51

Kim Stachenfeld

@neuro_kim

4 years

I've been working with Jraph internally for a while now and totally love it. Makes it super easy to put together new graph NN models. Congrats to colleagues on its release!

Thomas Kipf

@tkipf

4 years

Very excited about the release of Jraph. Finally an easy-to-use, extensive and fast Graph Neural Network library in JAX! Fully compatible with NN libraries such as Flax and Haiku:

14

178

1K

1

3

47

Kim Stachenfeld

@neuro_kim

4 years

"tell me about your research topic kim"

4

3

48

Kim Stachenfeld

@neuro_kim

4 years

DeepMind Internship applications open! Note they are accepted on a rolling basis, so apply now!

Google DeepMind

Artificial intelligence could be one of humanity’s most useful inventions. We research and build safe artificial intelligence systems. We're committed to solving intelligence, to advance science...

deepmind.google

0

21

44

Kim Stachenfeld

@neuro_kim

10 months

Come see @eckstein_maria and @kevinjmiller10 talk about classical cognitive modeling, data-driven modeling with neuroscience, and what lies in between! Then catch up with all three of us later for a hands-on tutorial where we learn to fit some models 😎

GitHub - kstach01/CogModelingRNNsTutorial

Contribute to kstach01/CogModelingRNNsTutorial development by creating an account on GitHub.

github.com

CogCompNeuro

@CogCompNeuro

10 months

Join us now for Keynote+Tutorials (8:30-10:15)! (1) A High-dimensional View of Computational Neuroscience (2) Cognitive models of Behavior: Classical and Deep Learning Approaches (3) Behind the Eyes: Insights into Cognition from Naturalistic Gaze Behavior in VR

1

2

22

2

7

45

Kim Stachenfeld

@neuro_kim

4 years

when i first saw this i thought it was an illustration of NN dropout

0

4

43

Kim Stachenfeld

@neuro_kim

7 years

Our new paper! Predictive representations in hippocampus:

2

24

42

Kim Stachenfeld

@neuro_kim

4 years

Stacy Abrams got 800,000 voters registered in GA over 6 years. She + her team are the reason the state is in play. We need 50 of her, 6 years ago.

Former Georgia Gubernatorial Candidate On A Push For Voter Turnout

NPR's Mary Louise Kelly talks with Stacey Abrams, former Georgia House minority leader and 2018 Democratic gubernatorial candidate, about efforts to get out and protect the vote in her state.

www.npr.org

Stacey Abrams

@staceyabrams

4 years

Here's where we are: 1. Local elections officials are taking the time necessary to make sure every eligible vote is counted. 2. Trump is losing and trying to crown himself the winner. 3. Voters choose the future. We fought for fairer elections and it’s working.

5K

18K

139K

3

6

43

Kim Stachenfeld

@neuro_kim

4 years

New paper out on interactions between striatum+hippocampus during RL! Check out thread by @jesse_geerts , and congrats to Fabian @NeilBurgess10 and @jesse_geerts !

0

4

43

Kim Stachenfeld

@neuro_kim

3 years

An opening on our team for a research scientist! Consider applying, especially if you like graphs, neural networks, or both at the same time.

Peter Battaglia

@PeterWBattaglia

3 years

We’re looking for a Research Scientist on our Structured Intelligence team at DeepMind - especially with experience or interest in GNNs. See the link below to apply, or feel free to reach out to me directly.

4

64

319

0

8

40

Kim Stachenfeld

@neuro_kim

2 years

But -- his over-the-top, theatrical, maximalist, recklessly passionate music and his performing persona are just my absolute favorite. Rest in peace, Meatloaf.

2

0

37

Kim Stachenfeld

@neuro_kim

5 years

@CompCogNeuro @behrenstimb If you want to explore some of these ideas with code, check out this jupyter notebook from the Dartmouth MIND 2019 Summer School. This github page contains a lot of great resources (slides+code) for thinking about cognitive maps. 2/3

2

6

34

Kim Stachenfeld

@neuro_kim

3 years

Whenever I write "we thank the reviewer for the thoughtful review", I feel a pang of sentimentality. I really am grateful for the thoughtful review.

0

34

Kim Stachenfeld

@neuro_kim

7 years

We made a paper!

bioRxiv Neuroscience

@biorxiv_neursci

7 years

The hippocampus as a predictive map #biorxiv_neursci

0

2

8

1

6

33

Kim Stachenfeld

@neuro_kim

3 years

Join us this evening on #LearmingSalon !

NeuroData Lab

@neuro_data

3 years

Very excited for @neuro_kim today at 4PM EST on #LearningSalon . While reading I was reminded of one of my favorite hippocampal papers , specifically, the Bayesian model with continuity constraints.

0

5

20

1

3

29

Kim Stachenfeld

@neuro_kim

4 years

if you're at #cosyne2020 , come check out poster I-20 in tonight's session by @jesse_geerts + @NeilBurgess10 + me! Keywords: hippocampus, probabilistic, prediction, structure learning, replay, and (of course) flexibility :)

0

2

30

Kim Stachenfeld

@neuro_kim

5 years

This is going to be fun :)

CogCompNeuro

@CogCompNeuro

5 years

Kim Stachenfeld ( @neuro_kim ) and Tim Behrens ( @behrenstimb ) will hold a Tutorial on "Representing states and spaces" at #CCN2019 ()

0

2

22

0

28

Kim Stachenfeld

@neuro_kim

2 years

To summarize our approach, we… 1. Use classical physics solvers to produce high-res trajectories 2. Downsample in space and time to produce training data 3. Train a neural network to do next-step prediction on low-res frames 4. Run it iteratively to generate long trajectories

1

3

27

Kim Stachenfeld

@neuro_kim

4 years

COUNTING not FINDING you maniac

Donald J. Trump

@realDonaldTrump

4 years

They are finding Biden votes all over the place — in Pennsylvania, Wisconsin, and Michigan. So bad for our Country!

143K

62K

467K

1

0

29

Kim Stachenfeld

@neuro_kim

4 years

It is a weird feeling to be sitting at home the night before a talk without having to give a talk the next day (videos were submitted in advance). Without slides to present tomorrow, I'm not sure what I should be making unnecessary changes to. Maybe I should cut my hair?

Kim Stachenfeld

@neuro_kim

4 years

Excited and honoured to be on a lineup with these badass researchers. #BAICS2020 ICLR workshop tomorrow (Sunday). Catch us from your couch.

0

4

18

1

0

27

Kim Stachenfeld

@neuro_kim

3 years

This was SO much fun. I really missed the part of research where you have fun after hours conversations about science. This was a perfect antidote to lockdown drear 🥰

Kim Stachenfeld

@neuro_kim

3 years

Join us this evening on #LearmingSalon !

1

3

29

5

2

27

Kim Stachenfeld

@neuro_kim

2 years

@AnnaSchapiro Work distinctive phrases from the response into casual conversation, watch for pupil dilation.

1

0

26

Kim Stachenfeld

@neuro_kim

1 year

It was SUCH an incredible honor to give a cosyne tutorial. Thanks so much for everyone who attended. In the spirit of learning from reinforcement, we'd love to get some feedback on what worked and what could be improved -- see this *very* short survey:

Cosyne 2023 RL Tutorial Feedback

We'd love to get feedback about the Cosyne 2023 RL Tutorial, both so we can improve the content of the RL lecture and so that we can optimize future Cosyne tutorials.

docs.google.com

Kim Stachenfeld

@neuro_kim

1 year

Tremendously excited to be giving an RL tutorial at #cosyne2023 ! Materials for the tutorial can be found here. slides coding colab (File > Make a copy; Runtime > Change runtime type > GPU)

5

30

147

1

3

23

Kim Stachenfeld

@neuro_kim

6 years

easy read + awesome review of hippocampal representations from katherine duncan + margaret schlichting

0

3

23

Kim Stachenfeld

@neuro_kim

2 years

In Baltimore for #ICML2022 -- come say hi!

1

0

23

Kim Stachenfeld

@neuro_kim

5 years

CCN paper on probabilistic Successor Representation with @jesse_geerts and @NeilBurgess10 ! Think SR + uncertainty + flexibility through non-local updates.

0

2

23

Kim Stachenfeld

@neuro_kim

2 years

"Now girls, tonight you are going to hear some very bad words. And I want you both to promise me, right now, that you are NEVER going to use these words at home." The audience laughs. Eyes wide, we nod dutifully to Meatloaf. "But... in school? It's perfectly okay."

1

0

21

Kim Stachenfeld

@neuro_kim

2 years

Turbulence simulation is vital to many areas of forecasting, engineering, and science. Simulators require high spatial and temporal resolution in order to make accurate predictions, as running at lower resolutions can lead to large errors because of chaotic dynamics.

2

1

20

Kim Stachenfeld

@neuro_kim

3 years

@criticalneuro @ChampalimaudF @blamlab Was fantastic!! My favorite action shot.

2

22

Kim Stachenfeld

@neuro_kim

3 years

Internship applications at DeepMind are now open! Applications are accepted on a rolling basis, so apply now :)

Google DeepMind

Artificial intelligence could be one of humanity’s most useful inventions. We research and build safe artificial intelligence systems. We're committed to solving intelligence, to advance science...

deepmind.google

1

8

20

Kim Stachenfeld

@neuro_kim

5 years

Tune in on or after June 26 to hear Andrew Saxe, Kristin Branson, Leila Wehbe, and me say things about ml+neuro at this virtual SfN conference! Register: Info:

Machine Learning in Neuroscience: Fundamentals and Possibilities

Machine learning methods enable researchers to discover statistical patterns in large datasets to solve a wide variety of tasks, including in neuroscience.

neuronline.sfn.org

Society for Neuroscience (SfN)

@SfNtweets

5 years

What is the future for the convergence of #machinelearning and #neuroscience ? The June 26 #SfNvirtual conference with Kristin Branson, Andrew Saxe, @neuro_kim , Leila Wehbe offers insights. Registration lets you watch any time up for 6 months afterwards:

2

10

13

2

5

21

Kim Stachenfeld

@neuro_kim

5 years

Made it to #ICML2019 in one piece. Let me know if you want to meet up! (Photo cred @criticalneuro )

1

0

20

Kim Stachenfeld

@neuro_kim

3 years

First in-person conference in almost 2 years!! Pretty amazing.

ChampalimaudResearch

@Neuro_CF

3 years

#CRS21 Live: @neuro_kim is discussing how #hippocampus and #entorhinalcortex build structured representations of the environment, which are used to support flexible reasoning. #placecells #gridcells #deeplearning

0

5

13

0

20

Kim Stachenfeld

@neuro_kim

4 years

I thought this work by esteemed colleagues @jhamrick and @shakir_za translating Marr's levels into ML was really interesting. Curious what other neuroscientists think.

Jess Hamrick is [email protected]

@jhamrick

4 years

You may remember a while back that I asked folks if they knew about Marr's Levels of Analysis. Many didn't, and I promised an explanation. At long last, I and @shakir_za are excited to share one, which will be presented at #BAICS2020 !

2

23

95

2

6

20

Kim Stachenfeld

@neuro_kim

4 years

AI storytime with @FeryalMP ! particularly looking forward to checking out chelsea finn's metalearning lectures and david mackay's information theory lectures.

Google DeepMind

@GoogleDeepMind

4 years

Today’s suggestions are from research scientist Feryal Behbahani @feryalmp ! #AtHomeWithAI

2

23

252

0

3

20

Kim Stachenfeld

@neuro_kim

3 years

Seen here enjoying my margarita after 3 hours of #LearningSalon

0

18

Kim Stachenfeld

@neuro_kim

3 years

@behrenstimb A British colleague and I once wondered why a group of ~20 unfamiliar people in our building were wearing matching raincoats and rainboots. I decided to go over and ask. He was horrified. I love being an American in England.

2

0

18

Kim Stachenfeld

@neuro_kim

2 years

If you're at #cosyne2022 , check out @NotGeorgeTom 's poster in tonight's session! Hippocampally plausible learning of predictive maps 😎

Will de Cothi

@willdecothi

2 years

#cosyne2022 come check out the insanely talented @NotGeorgeTom present our poster I-050 8pm tonight (17th) on learning predictive representations using STDP and phase precession! Work done with the awesome @neuro_kim & @caswellcaswell

0

1

34

0

3

18

Kim Stachenfeld

@neuro_kim

4 years

Excited and honoured to be on a lineup with these badass researchers. #BAICS2020 ICLR workshop tomorrow (Sunday). Catch us from your couch.

Bridging AI and Cognitive Science #BAICS2020

@baicsworkshop

4 years

In the days leading up to #BAICS2020 , we'll be tweeting about our incredible speakers and panelists!

1

4

23

0

4

18

Kim Stachenfeld

@neuro_kim

3 years

😂😭 nice article on representational drift though

Justin Kiggins

@neuromusic

3 years

Only one?

6

27

326

1

0

18

Kim Stachenfeld

@neuro_kim

6 years

Neat work unpacking transitive inference mechanism in hippocampus, using fancy schmancy fmri methods, from deepmind neuro + u magdeburg peops! Congrats :)

Google DeepMind

@GoogleDeepMind

6 years

Our latest @NeuronCellPress paper, in collaboration with @etude_yi & @david_berron from @DuzelLab , describes how our brains derive new insights by drawing on multiple related experiences connected by dynamic links within a network of hippocampal memories

3

124

269

0

4

17

Kim Stachenfeld

@neuro_kim

1 year

A Comp Neuro & ML course is happening in Ghana this year, with input from researchers at Janelia and DM! Seems cool :)

TReND in Africa

@TReNDinAfrica

1 year

Apply now for our "Computational Neuroscience & Machine Learning" course Shout-out to the organizers @saraysoldado @jsoldadomagrane @gabymychel @Tom_DAT @drpatamot @Duutabib @esztervertes @artemiskoum and @Deepmind & @HHMIJanelia for their contributions🤩

1

66

110

0

2

17

Kim Stachenfeld

@neuro_kim

6 years

solid reading list on AI + cognitive science

0

3

18

Kim Stachenfeld

@neuro_kim

2 years

I'll be talking today (1:30ET Ballroom 2)in #ICML2022 Beyond Bayes workshop about using learned simulators for general purpose design! Work with @KelseyRAllen @ZePoLiTaT Alvaro Sanchez-Gonzalez @PeterWBattaglia @jhamrick @spectralhippo .

Emily Mackevicius

@e_mackevicius

2 years

🧩💡Co-organizing #ICML2022 Beyond Bayes workshop on reasoning tomorrow, w/ @ZennaTavares et. al.💡🧩 I'll moderate a panel w/ @neuro_kim , Ishita Dasgupta, & @tylerraye ! Speakers include @rosemary_ke , @sugsharma , @chingfang17 , @guyvdb , @jwvdm ! Schedule:

0

10

24

0

3

17

Kim Stachenfeld

@neuro_kim

2 years

We use a “domain-general” architecture comprised of CNN modules that are not specific to turbulence simulation. It consists of a linear CNN Encoder and Decoder, as well as processors comprised of a U-shaped stack of dilated CNNs and residual skip connections.

1

15

Kim Stachenfeld

@neuro_kim

5 years

Come check out this work on graph-structured agents that build things, and see some cool videos of agents playing with blocks! 👷‍♀️🤖🏗️ Also just now realizing we may have missed an opportunity to name this something with the acronym BOB (as in, the builder). Alas.

Google DeepMind

@GoogleDeepMind

5 years

How do you get an agent to build complex structures out of blocks? Come see our answer on Wednesday at #ICML2019 from 11:40am-12pm in Hall B & during posters! We use 3 key ingredients: structured policies, object-centric actions, & model-based planning.

1

28

91

1

7

16

Kim Stachenfeld

@neuro_kim

2 years

Check out the poster tonight 18:30-20:30 PDT (ICLR Session 9, Room 2, Spot C1)! Also check out... ICLR virtual poster: Paper: Additional videos:

Learned Coarse Models for Efficient Turbulence Simulation

Examples of the Dil-ResNet trained on different domains, capturing a range of qualitatively diverse turbulent dynamics. (Please see the menu for additional comparisons).

sites.google.com

2

3

16

Kim Stachenfeld

@neuro_kim

5 years

This is going to be awesome! Looking forward to participating. If you're a grad student interesting in learning neuro methods related to cognitive maps, apply!

Methods in Neuroscience at Dartmouth (MIND)

@comp_mind_ss

5 years

We are now accepting applications for the 3rd annual Methods in Neuroscience at Dartmouth (MIND) Computational Summer School (Due 4/15). This year we will be focusing on cognitive maps from the perspectives of systems, cognitive, & social neuroscience

4

46

43

0

2

15

Kim Stachenfeld

@neuro_kim

4 years

Hurray for Neuroscience Without Borders! Fighting the good fight for data standardization with a #cosyne2020 tutorial

0

1

13

Kim Stachenfeld

@neuro_kim

2 years

We also find that the learned model outperforms a state-of-the-art classical turbulence simulator, Athena++, at a comparably low-resolution, in terms of pixel-wise loss (blue v. orange). An intermediate resolution Athena++ (green) beats both.

1

14

Kim Stachenfeld

@neuro_kim

2 years

Stability can be hard for long rollouts: small errors accumulate, which can eventually lead to domain-shift. Adding noise to the inputs during training can lead to dramatically better stability (although as shown below, this is less of a problem for large timesteps).

1

14

Kim Stachenfeld

@neuro_kim

4 years

YAY, grid cells in a rat jungle gym! Cannot wait to read.

0

1

13

Kim Stachenfeld

@neuro_kim

2 years

However, where the learned model really shines is preserving high-frequency information: spectral error is lower than Athena++ at the low and intermediate resolution, as the learned model maintains the sharpness of certain features Athena++ smooths away.

1

2

13

Kim Stachenfeld

@neuro_kim

5 years

Really excited to be on the @DeepMindAI panel at the @CheltSciFest !

Murray Shanahan

@mpshanahan

5 years

The @CheltSciFest programme is out. I'm on a @DeepMindAI panel with @FryRsquared @neuro_kim and @arkitus #cheltscifest

0

3

21

0

13

Kim Stachenfeld

@neuro_kim

7 years

MIT tech review wrote a post about our recent paper!

MIT Technology Review

@techreview

7 years

DeepMind proposes that the human hippocampus makes predictions about the future, and AIs should mimic it.

5

55

83

0

3

14

Kim Stachenfeld

@neuro_kim

2 years

Our simple general-purpose learned models quantitatively outperform more specialized, parameterized models. However, most learned models do qualitatively pretty well, suggesting learned models are generally a good fit for this type of problem.

1

2

13

Kim Stachenfeld

@neuro_kim

5 years

Thanks to the other panelists and Hannah for the wonderful discussion and to the audience at @CheltSciFest for the brilliant questions! What a delightful experience!

Google DeepMind

@GoogleDeepMind

5 years

If you’re @CheltSciFest , join us at 6.30pm today for What is AI anyway? Our panelists @arkitus , @neuro_kim & @mpshanahan , and moderator @FryRsquared , will take a trip through the world of AI and discover where it might go in the future #whatisAI #CheltSciFest

2

8

16

0

2

12

Kim Stachenfeld

@neuro_kim

4 years

@criticalneuro alternate No: I already talked to myself a lot.

0

13

Kim Stachenfeld

@neuro_kim

3 years

it is weird to circle back to memorable events that coincided with the start of the pandemic. for me that is cosyne and asparagus being back in season.

Alex Piet

@alexpiet

3 years

For me #cosyne21 marks the one year anniversary of COVID. When I left for #cosyne20 the world was normal. During the conference covid changed from being a news story to being all around me. By the time I returned to Seattle, the city was on lockdown and work was remote. 1/3

3

0

21

0

12

Kim Stachenfeld

@neuro_kim

2 years

Accelerating turbulence simulation is of particular interest to astrophysics, which requires very large, very high-resolution grids to simulate turbulent phenomena, such as eg interstellar medium.

1

0

12

Kim Stachenfeld

@neuro_kim

2 years

Generalizing to different box sizes for “Mixing Layer” turbulence is also tricky. Training on a range of box sizes improves predictions on an even larger range. But artifacts persist, and quantitatively, the predictions of both models remain outside of the expected error.

2

12

Kim Stachenfeld

@neuro_kim

4 years

@behrenstimb my dad used to offer awards for outrunning him. he was not very fast but he would tell good stories, so when i was thoroughly absorbed he would shove me into shrubbery and sprint off. i imagine this strategy could generalize to mario kart.

1

0

12

Kim Stachenfeld

@neuro_kim

3 years

without produce from the EU we are now left with... cabbage. and rhubarb. send recipes ☹️

How dependent is the UK on the EU for food?

France's ban on freight from Britain caused havoc for lorry drivers, but will it affect shoppers?

www.bbc.co.uk

1

0

12

Kim Stachenfeld

@neuro_kim

2 years

If you are at #cosyne2022 , collaborator Gabriela Michel (not on Twitter) is presenting work tonight from her previous lab modeling cortical development across species. Check it out, it's super cool!

1

12

Kim Stachenfeld

@neuro_kim

5 years

Generative Models + Model-based RL in Robotics + AI today in Hall A at #ICML2019 ! Stellar lineup: @ylecun @jhamrick Stefan Schaal, David Silver, @ByronBoots @chelseabfinn Abhinav Gupta #ICMLmodeling

0

5

12

Kim Stachenfeld

@neuro_kim

5 years

I love this picture

Alexander Laurence 👨‍💻

@AIexLaurence

5 years

Understanding properties of learning in an artificial context can translate to insights about learning in the biological context, and discuss how machine learning problems and unexplained neuroscience data can mutually inform each other... ( @neuro_kim and Andrew Saxe at #SfN19 )

2

0

10

0

2

11

Kim Stachenfeld

@neuro_kim

3 years

Hey everyone! Reminder that internship applications for this round close Oct 4. There will be another window Dec 6-17 too.

Kim Stachenfeld

@neuro_kim

3 years

Internship applications at DeepMind are now open! Applications are accepted on a rolling basis, so apply now :)

1

8

20

0

5

10

Kim Stachenfeld

@neuro_kim

5 years

Cool new successor feature work!

Steven Hansen

@Zergylord

5 years

Excited to share some new work on ArXiv today: "Fast Task Inference with Variational Intrinsic Successor Features" Done with @wwdabney , Andre Barreto, Tom Van de Wiele, @dwf , and @VladMnih TL;DR: Unsupervised pre-training for efficient RL

4

38

153

0

10

Kim Stachenfeld

@neuro_kim

6 years

Late to the distill party but these are so so so beautiful

0

2

10

Kim Stachenfeld

@neuro_kim

2 years

Meatloaf died, probably of covid, possibly unvaccinated, within a year of his musical partner, Jim Steinman, who wrote Meatloaf's hits (also others like "Total Eclipse of the Heart"). I didn't have a lot of ideological overlap with Meatloaf (as "possibly unvaccinated" suggests).

1

0

11

Kim Stachenfeld

@neuro_kim

2 years

This work was an interdisciplinary collaboration between researchers at @deepmind , @FlatironCCA , @googIeresearch and Princeton University. This was a wonderful experience, and involved fascinating conversations about the fluid dynamics of outer space.

1

11

Kim Stachenfeld

@neuro_kim

2 years

Since I was little, my family has been obsessed with Meatloaf. When my sister and I were little, we would bounce around the house, enthusiastically belting out the lyrics to Paradise by the Dashboard Light, understanding about 8% of it (why won't she just let him sleep?) (2/n)

2

0

10

Kim Stachenfeld

@neuro_kim

2 years

When we were young (maybe 8 and 10?) my parents took us to a Meatloaf concert. This was a huge deal. It was a 10pm on a school night, it was in New York City, and we had seats pretty close to the stage. And it was Meatloaf!

1

0

11

Kim Stachenfeld

@neuro_kim

5 years

the hero we need not the hero we deserve, thanks @dabelcs

David Abel

@dabelcs

5 years

Notes from @icmlconf here: #ICML2019

18

213

828

0

3

11

Kim Stachenfeld

@neuro_kim

4 years

i did not adequately predict how much this would make me feel feelings now someone get this woman a glass of chardonnay AND A DEMOCRATIC SENATE!

1

0

10

Kim Stachenfeld

@neuro_kim

2 years

We find that the same architecture can accurately capture diverse types of turbulent and unstable dynamics at low spatial+temporal resolutions!

1

9

Kim Stachenfeld

@neuro_kim

4 years

great graphic

Iris van Rooij 💭

@IrisVanRooij

4 years

This puts some things in perspective

6

42

173

0

10

Kim Stachenfeld

@neuro_kim

2 years

@ZePoLiTaT @KelseyRAllen As well as @ZePoLiTaT and @KelseyRAllen , this work was a collaboration with @spectralhippo @jhamrick @PeterWBattaglia and Alvaro Sanchez-Gonzalez. Working with this brilliant group, and especially co-leading a project with Tatiana+Kelsey, has been sheer delight.

1

0

10

Kim Stachenfeld

@neuro_kim

5 years

Amazing tour de force of a talk on graphnet agents that build, @jhamrick powered through a briefly broken projector like a badass without missing a step, and by faaaar the most fashionable headwear of ICML 👷‍♀️ check out poster #36 tonight for more amazingness #ICML2019

Jess Hamrick is [email protected]

@jhamrick

5 years

Excited to share our newest work (& my first final-author paper!) led by Victor Bapst and Alvaro Sanchez-Gonzalez (w/ @CarlDoersch @neuro_kim @pushmeet @PeterWBattaglia ). We investigate structured, object-centric agents that can build things!👷‍♀️🏗

6

35

135

1

2

10

Kim Stachenfeld

@neuro_kim

4 years

@pfau @jhamrick And sign up to volunteer virtually with the New Georgia Project!

1

4

10

Kim Stachenfeld

@neuro_kim

3 years

One time the only doctor on an Antarctic expedition removed his own appendix when he came down with appendicitis. Now these scientists produce a paper on long covid while actively having long covid! Kudos.

The man who cut out his own appendix

Russian surgeon, Leonid Rogozov, got appendicitis while on an Antarctic mission - he had to operate on himself.

www.bbc.co.uk

Athena Akrami

@AthenaAkrami

3 years

And here is the preprint: "Characterizing Long COVID in an International Cohort: 7 Months of Symptoms and Their Impact" Special thanks to thousands of you who participated in our study on #LongCovid . a thread 🧵 0/

40

321

722

0

9

Kim Stachenfeld

@neuro_kim

3 years

i love this question from physics stack exchange

Why can't the human eye focus to make blurry photos/video clear?

The human eye focuses by flexing the lens, changing its focal length. When we switch from looking at a near object to a far object, our lens flexes, moving the focal length such that the near obje...

physics.stackexchange.com

1

10

Kim Stachenfeld

@neuro_kim

4 years

@AthenaAkrami @criticalneuro I've been recently taken in with how cognitive immune systems are... Even without the neuro prefix, they have memory, recognition, learning, search. Cryptographically secure protocols! They have biases and misunderstandings that seem almost like mental illness.

3

0

10