Kim Stachenfeld

@neuro_kim

Research scientist at DeepMind. Likes hippocampi and learning and other things brains have and do, too.

Vrijeme pridruživanja: rujan 2016.

Tweetovi

Blokirali ste korisnika/cu @neuro_kim

Jeste li sigurni da želite vidjeti te tweetove? Time nećete deblokirati korisnika/cu @neuro_kim

  1. Prikvačeni tweet
    13. ruj 2019.

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

    Prikaži ovu nit
    Poništi
  2. 14. lis 2019.

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

    Poništi
  3. 18. ruj 2019.

    Great thread on sources for intuitive explanations of math/CS things.

    Poništi
  4. proslijedio/la je Tweet
    15. ruj 2019.

    Presenting a poster at this afternoon on how to incorporate biological details into deep nets. Certainly on the prettier side of posters I normally make...

    Prikaži ovu nit
    Poništi
  5. proslijedio/la je Tweet
    22. kol 2019.

    Hippocampal Remapping as Hidden State Inference joint work with and Matt Wilson

    Poništi
  6. 13. ruj 2019.
    Prikaži ovu nit
    Poništi
  7. 13. ruj 2019.

    Obligatory caveats like: we made these slides over a small number of days / we tried but they're not a comprehensive survey / certainly biased towards familiar work. If you see something wrong or missing, message and i'll augment the deck 3/3

    Prikaži ovu nit
    Poništi
  8. 13. ruj 2019.

    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

    Prikaži ovu nit
    Poništi
  9. proslijedio/la je Tweet
    13. ruj 2019.

    Kim introduces "PI life MDP" (from Andrew Moore?) in tutorial on states and spaces by Kim stachenfeld and Tim behrens

    Poništi
  10. proslijedio/la je Tweet
    9. kol 2019.

    Happy to announce our Meta-learning workshop call for papers! Due Sept 10 at midnight, AoE Check out details and our stellar lineup of speakers at Submission link:

    Poništi
  11. proslijedio/la je Tweet
    28. lip 2019.

    We reduced the training time of a sparse graph neural net from 1 day to 13 mins (!) on TPU with large batch training. Key is identifying band-diagonal structure in the adjacency matrix. Work with @subho87 Paper

    Poništi
  12. proslijedio/la je Tweet
    28. lip 2019.

    We're looking for subjects (18-60 yrs old, male or female) to participate in our Human Psychophysics experiments - Auditory perception, memory & decision making! Experiment takes 60-90 min/ses, possibly multiple sessions, at . Read more: RT pls!

    Poništi
  13. proslijedio/la je Tweet
    6. svi 2019.

    New in Nature Neuro from Eero Simoncelli's group: Perceptual straightening of natural videos

    Poništi
  14. proslijedio/la je Tweet
    26. lip 2019.
    Poništi
  15. proslijedio/la je Tweet

    One of the ways we can improve representation in science is by tackling the wild beast that is Wikipedia. is hosting this Weds 26th June, so if you're down for a fun, free and meaningful event, check it out!

    Poništi
  16. proslijedio/la je Tweet
    25. lip 2019.

    Congratulations to who is featured in ’s list for her work on biological reinforcement learning and the hippocampal formation

    Poništi
  17. 26. lip 2019.

    Utterly honored to be included on this list with all these super cool people. Thank you

    Poništi
  18. 25. lip 2019.
    Poništi
  19. proslijedio/la je Tweet
    18. lip 2019.

    We updated our on memory distortions through boundary geometry: We use the successor representation to show how distorted grid patterns can deform . Work with , , T Ruiter, &

    Poništi
  20. 19. lip 2019.

    Heyo, distorted successor representation grids! Pretty cool.

    Poništi

Čini se da učitavanje traje već neko vrijeme.

Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.

    Možda bi vam se svidjelo i ovo:

    ·