Thomas Mesnard

@Mesnard_Thomas

Research Assistant at DeepMind and PhD student at École Polytechnique

Vrijeme pridruživanja: listopad 2015.

Tweetovi

Blokirali ste korisnika/cu @Mesnard_Thomas

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

  1. proslijedio/la je Tweet
    5. pro 2019.

    Really excited for next week and to present our spotlight on credit assignment :) tl;dr We can rewrite value functions in terms of a hindsight quantity that explicitly captures credit assignment and get a whole new family of RL algs! 🥳

    Prikaži ovu nit
    Poništi
  2. proslijedio/la je Tweet
    30. sij 2019.

    Paper: replay extends for hours - very nice to see this result, it's something that's bugged me for years. We talk about replay and consolidation during sleep, but only observe replay in rodents soon after an experience. Good to see this gap filled!

    Poništi
  3. 16. pro 2018.

    Best thing you will see today!

    Poništi
  4. proslijedio/la je Tweet
    15. pro 2018.

    Since reading this 6 months ago, I've come to think it's half a dozen of the best paragraphs I've ever read on how to get much, much better at anything: (by @autotrnslucence )

    Prikaži ovu nit
    Poništi
  5. proslijedio/la je Tweet
    6. pro 2018.

    Happy birthday to Geoffrey Hinton, whose work in neural networks has led many to call him the “Godfather of AI”: (credit: )

    Poništi
  6. proslijedio/la je Tweet
    6. pro 2018.

    The full peer-reviewed evaluation of is here - a single algorithm that creatively masters chess, shogi and Go through self-play

    Prikaži ovu nit
    Poništi
  7. proslijedio/la je Tweet
    30. stu 2018.

    Interesting read on disentangled representation learning by Locatello et al. They argue that "unsupervised learning of disentangled representations is fundamentally impossible without inductive biases", and that these biases should be made explicit

    Poništi
  8. 25. stu 2018.
    Poništi
  9. proslijedio/la je Tweet
    24. lis 2018.

    Check out new research from the Google Brain team, and ETH Zürich which allows RL agents to explore environments in an intelligent way — and avoid undesired behaviours stemming from surprise-based curiosity. Learn more on the Google AI blog at

    Poništi
  10. 10. lis 2018.

    Hey folks! I'm looking for a place to live for the next few months not too far away from King's Cross Station, London. If you know someone who is looking for a roommate/tenant or if you have any good tips, please let me know!

    Poništi
  11. 10. lis 2018.

    Great news! I'll be in London for the next few months for an internship at Google DeepMind. Super excited to start this new adventure! Let me know if you happen to find yourself in London during that time 😀!

    Poništi
  12. proslijedio/la je Tweet
    13. ruj 2018.

    Blake Richards regales me with his beautiful recent work exploring how deep learning and credit assignment could work in the brain... And he really, really doesn't like zombies

    Poništi
  13. proslijedio/la je Tweet
    8. ruj 2018.

    Thanks to and everyone else who came out at to our breakout session on systems consolidation. It was a really thought provoking discussion!!!!

    Poništi
  14. proslijedio/la je Tweet
    8. ruj 2018.

    How does the cortex solve the credit assignment problem? and Tim Lillicrap suggest that the cortex has figured it out by having separate compartments for feedforward and feedback information flow in the pyramidal neurons. More data needed!

    Poništi
  15. proslijedio/la je Tweet
    6. ruj 2018.

    This is a really important paper, IMO. If you’re interested in neocortical learning, you really should read this!

    Poništi
  16. 6. ruj 2018.

    Great work! You should definitely check this out!

    Poništi
  17. proslijedio/la je Tweet
    4. ruj 2018.

    The main conference sold out in 11 minutes 38 seconds

    Poništi
  18. proslijedio/la je Tweet
    15. kol 2018.

    My and Tim Lillicrap’s correspondance in NRN with and on their recent review on cortical deep learning is now available!

    Prikaži ovu nit
    Poništi
  19. proslijedio/la je Tweet
    15. kol 2018.

    Our new paper, which generalizes Equillibrium propagation to general vector field dynamics is finally out! Joint work with Benjamin Scellier, Jonathan Binas, Thomas Mesnard and Yoshua Bengio! .

    Poništi
  20. proslijedio/la je Tweet
    9. kol 2018.

    1/ Who wants to hear some scientific intrigue? A few weeks ago, a group of physical chemists posted a paper online announcing the observation of superconductivity at room temperature. Today I posted a comment pointing out something funny in their data.

    Prikaži ovu nit
    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:

    ·