Stephen McAleer

@McaleerStephen

PhD student in deep reinforcement learning at UC Irvine

Vrijeme pridruživanja: srpanj 2014.

Tweetovi

Blokirali ste korisnika/cu @McaleerStephen

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

  1. proslijedio/la je Tweet
    25. sij

    This semester I'm teaching a new PhD course "Economics, AI, and Optimization." I'll be covering how AI/Opt methods enable large-scale economic solution concepts. I'm planning to share lectures notes that I hope will be of broader interest.

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

    AI and economics-style utility true believers should really read John Dewey (, thinking of you!). He really makes so clear how incoherent a theory of preference is outside of a theory of sociality and communication (hat tip to on this).

    Prikaži ovu nit
    Poništi
  3. proslijedio/la je Tweet
    16. sij

    Neural Replicator Dynamics: bringing replicator dynamics to a new level in reinforcement learning - this will unlock some more exciting work in the future! "Neural Replicator Dynamics: Multiagent Learning via Hedging Policy Gradients"

    Prikaži ovu nit
    Poništi
  4. proslijedio/la je Tweet
    31. lis 2019.

    Meta Reinforcement Learning is good at adaptation to very similar environments. But can we meta-learn general RL algorithms? Our new approach MetaGenRL is able to. With and Paper: Blog:

    Poništi
  5. proslijedio/la je Tweet
    16. lis 2019.

    Excited to share that we are organizing the AAAI Workshop on Reinforcement Learning in Games (AAAI-RLG), to be held February 7th or 8th in New York City! Submissions due November 15th. with Julien Perolat and Marc Lanctot

    Poništi
  6. proslijedio/la je Tweet
    27. ruj 2019.

    RLlib callbacks are a gamechanger: storing custom env metrics directly in tensorboard is invaluable.

    Poništi
  7. proslijedio/la je Tweet
    30. kol 2019.

    Someone is obviously really close to solving AGI:

    Poništi
  8. proslijedio/la je Tweet
    30. kol 2019.

    Our paper on the poker AI is now in the print edition of , and we're on the front cover!

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

    We're excited to release OpenSpiel: a framework for reinforcement learning in games. It contains over 25 games, and 20 algorithms, including tools for visualisation and evaluation. GitHub: Paper:

    Poništi
  10. 14. kol 2019.

    This is a great article about the problems with the attention economy: What is the Price of our Attention? by Quentin LE GARREC

    Poništi
  11. 14. kol 2019.

    This is what happens when your objective is to maximize users screen time. YouTube needs to drastically change their recommendation system.

    Poništi
  12. 15. srp 2019.

    Our paper "Solving the Rubik's Cube with Deep Reinforcement Learning and Search" has been published in Nature Machine Intelligence. You can check it out here:

    Poništi
  13. proslijedio/la je Tweet
    10. lip 2019.

    We've compiled a meta-reading list for our meta-learning tutorial: Short list of the main papers we covered in our meta-learning tutorial:

    Poništi
  14. 6. svi 2019.

    Excited to be at to present our paper on solving the Rubik's cube with reinforcement learning.

    Poništi
  15. proslijedio/la je Tweet
    20. ožu 2019.

    PEARL: Meta-RL that is 20-100x faster than prior methods, with better final performance, using soft actor-critic and order-invariant context embedding: w/ K. Rakelly, A. Zhou, D. Quillen, (ours is the blue one)

    Poništi
  16. proslijedio/la je Tweet
    1. ožu 2019.

    Feedback loops in recommendation systems can give rise to “echo chambers” and “filter bubbles” which can narrow a user’s content exposure, and ultimately shift their world view.

    Prikaži ovu nit
    Poništi
  17. proslijedio/la je Tweet
    5. ožu 2019.

    We introduce α-Rank, a principled method to evaluate multi-agent strategies, grounded in a new game-theoretic solution concept, Markov-Conley Chains, unique & tractable to compute. Joint work @karltuyls, S. Omidshafiei, C. Papadimitriou and G. Piliouras:

    Poništi
  18. proslijedio/la je Tweet
    22. velj 2019.

    For folks looking for a thorough intro to the mathematical foundations of reinforcement learning: Video lectures for Bertsekas’ course on RL and control are now available here:

    Poništi
  19. 17. velj 2019.

    For example, the main goal of AI research seems to be to create human-level intelligence. Is that really something that we want? Although seems relatively harmless, it's good to start thinking about what is worth researching in the first place.

    Prikaži ovu nit
    Poništi
  20. 17. velj 2019.

    I agree with many of these points. However, instead of doing research on potentially harmful technology and then not releasing it, the research community needs to have a discussion about what research is not worth doing in the first place.

    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:

    ·