𝚠𝚒𝚕𝚕𝚒𝚊𝚖  

@wgussml

I work on deep reinforcement learning and deep learning theory. PhD student at Carnegie Mellon University.

Pittsburgh, PA
Vrijeme pridruživanja: prosinac 2017.

Tweetovi

Blokirali ste korisnika/cu @wgussml

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

  1. Prikvačeni tweet

    I am really happy to announce that MineRL has been featured on Nature news! This means so much to us, after nearly two years of work w/o much external feedback! w/ & more!!

    Poništi
  2. proslijedio/la je Tweet
    2. velj

    Got promoted to a full professor at CMU:) Thanks to my amazing students, collaborators, and my mentors. CMU is a remarkable place, but what makes it truly unique is its brilliant students! Hard to find a better place to do ML and AI.

    Poništi
  3. Hey Berkeley ML folks, where is the new building!? SDH is a ghost town!

    Poništi
  4. proslijedio/la je Tweet
    21. sij

    Learnability (VC dim) is a *topological property*, as I proved in for parity, conjunctions, poly threshold fctns. Now this extends to downward-closed classes, conjunction of parities, and k-CNFs, as well! Just how far does this go?

    Prikaži ovu nit
    Poništi
  5. proslijedio/la je Tweet
    7. sij
    Odgovor korisniku/ci

    What you've said is 100% right. We don't measure it. And we've also forgotten what the real point of reviewing out to be: to advance science as rapidly as possible. This problem has technological, sociological, etc. aspects.

    Poništi
  6. proslijedio/la je Tweet
    30. pro 2019.
    Poništi
  7. I am curious to hear everyone's thoughts and responses to his claim!

    Prikaži ovu nit
    Poništi
  8. Here's 's response to the NIPS vs NeurIPS change. As mentioned, implies this type of censorship is against the grain of intersectionality, contributing to the sexualization of the word/not condemning those making the sexist insults.

    Prikaži ovu nit
    Poništi
  9. proslijedio/la je Tweet
    24. pro 2019.

    What the hell is going on with ML paper publishing?

    Prikaži ovu nit
    Poništi
  10. proslijedio/la je Tweet
    15. pro 2019.

    Minecraftを用いた大規模データセット及び強化学習研究環境「MineRL」の紹介。データの解説のみにとどまらず、既存手法での実験結果とその考察まで行っています。現実空間に近い仮想環境で、いかに効率よく学習させるかが鍵となる今、強化学習の発展に関わる注目の記事です。

    Poništi
  11. proslijedio/la je Tweet

    Day three of Dagstuhl is in full swing. With talks this morning including from talking about the MineRL Minecraft project.

    Prikaži ovu nit
    Poništi
  12. proslijedio/la je Tweet
    17. pro 2019.

    If you thought finding a diamond in MineRL was a hard AI challenge... try finding a Christmas gift for my dad

    Prikaži ovu nit
    Poništi
  13. Huge amount of equal contribution with and Nicholay Topin! I wish we could all present together!

    Prikaži ovu nit
    Poništi
  14. Hey ! If you missed the competition track this morning, I'll be giving a talk in the Deep RL Workshop on our Diamond Competition today! We've been stunned with the progress people have made on Minecraft in the low sample regime! 👉👉 4:45 PM West Expo C 👈👈

    Prikaži ovu nit
    Poništi
  15. proslijedio/la je Tweet
    14. pro 2019.

    session in Competition Track is kicked-off by . Winners are announced here. Don't miss it. Come to 116+117 West

    Poništi
  16. A wonderful thread on the competition! We had some absolutely amazing results. Stay tuned for videos of the agents and a follow up!

    Poništi
  17. proslijedio/la je Tweet
    13. pro 2019.

    Stop by our poster "Zero-shot Text Classification With Generative Language Models" today at the Meta-Learning workshop. Learn about the intersection of Meta Learning/NLP and some of the challenges going forward. paper:

    Prikaži ovu nit
    Poništi
  18. Wow, I am so excited! The competition was featured in The Verge! We've come a long way towards Minecraft AI in the past 2 years, but this competition is a great reminder that there are so many more challenges to solve! Read More:

    Poništi
  19. proslijedio/la je Tweet
    12. pro 2019.

    For saddle point problems and more general games, gradient descent can cycle and diverge. Using an extrapolation steps adresses this issue and converges to a saddle point (Nash equilibrium).

    Poništi
  20. proslijedio/la je Tweet
    3. pro 2019.

    On Dec. 12, a unique competition that challenges AI to discover virtual diamonds in the game Minecraft will publish the final results from the top 10 teams. More details in my story for .

    Prikaži ovu nit
    Poništi
  21. proslijedio/la je Tweet
    8. pro 2019.

    Excited to attend my first this year, thanks to travel grants from and ! Looking forward to presenting my work and learning from the vast ocean of amazing work and talented people present there!

    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:

    ·