Theo Weber

@theophaneweber

Research scientist at DeepMind AI/ML - probabilistic modeling, variational inference, reinforcement learning, deep learning

Vrijeme pridruživanja: srpanj 2007.

Tweetovi

Blokirali ste korisnika/cu @theophaneweber

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

  1. proslijedio/la je Tweet
    9. sij

    We are organizing a workshop on Causal learning for Decision Making at along with , Jovana Mitrovic, , Stefan and . Consider submitting your work!

    Poništi
  2. proslijedio/la je Tweet
    6. pro 2019.

    Check out our extensive review paper on normalizing flows! This paper is the product of years of thinking about flows: it contains everything we know about them, and many new insights. With , , , . Thread 👇

    Prikaži ovu nit
    Poništi
  3. proslijedio/la je Tweet
    30. lis 2019.

    My PhD thesis is now available on arXiv: Neural Density Estimation and Likelihood-free Inference There's a lot in it for those interested in probabilistic modelling with normalizing flows, and in likelihood-free inference using machine learning. (cont.)

    Prikaži ovu nit
    Poništi
  4. 16. lis 2019.

    Really excited to share this new paper with Lars Buesing and Nicolas Heess: By following connections between RL and inference, we inspire ourselves from alpha-zero style MCTS (for optimization) and develop a tree-search based alg. for inference.

    Prikaži ovu nit
    Poništi
  5. proslijedio/la je Tweet
    16. lis 2019.

    Great new paper by Lars, Nicholas and expanding the tools we have for doing approximate inference in discrete probabilistic models using MCTS 🤩 We need more such papers connecting our understanding of RL and inference to important models

    Poništi
  6. proslijedio/la je Tweet
    10. lis 2019.

    Wanna play around with SPIRAL but the installation seems complicated? I've just built a Docker image to make the experience as hassle-free as possible. To get the agent up and running on your machine follow the instructions here: Have fun!

    Prikaži ovu nit
    Poništi
  7. proslijedio/la je Tweet
    6. ruj 2019.

    Thrilled to be able to share what I've been working on for the last year - solving the fundamental equations of quantum mechanics with deep learning!

    Prikaži ovu nit
    Poništi
  8. proslijedio/la je Tweet
    13. kol 2019.

    We are excited to release Behaviour Suite for Reinforcement Learning, or ‘bsuite’ – a collection of carefully-designed experiments that investigate core capabilities of RL agents GitHub: Paper:

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

    Really excited to release to the public! - Clear, scalable experiments that test core capabilities. - Works with OpenAI gym, Dopamine. - Detailed colab analysis - Automated LaTeX appendix Example report:

    Poništi
  10. proslijedio/la je Tweet

    I'm developing a pet peeve around slides showing children learning things "one/few-shot", allegedly super magically. A child does not have a few months/years of experience. It has about 500 million years of experience.

    Poništi
  11. 10. sij 2019.

    I'm sure I am missing many references so if you believe I missed something please don't hesitate to ping me! (Similarly if you see anything dodgy mathematically)

    Prikaži ovu nit
    Poništi
  12. 10. sij 2019.

    And nevermind on the second paper - it will be on arxiv soon enough :)

    Prikaži ovu nit
    Poništi
  13. 10. sij 2019.

    Also recently published two papers: TD-VAE (), with Karol Gregor, , and friends. How can an agent build a temporally abstract model of the world, and use it to compute a 'belief state' - a representation of the agent's uncertainty.(Oral at ICLR)

    Prikaži ovu nit
    Poništi
  14. 10. sij 2019.

    Helps to connect and understand current algorithms, and hopefully offers a turnkey methodology to derive new ones in increasingly structured models. Find out more at , or come chat during AISTATS! Feedback very welcome :)

    Prikaži ovu nit
    Poništi
  15. 10. sij 2019.

    New paper out! It deals with credit assignment in stochastic computation graphs. Attempts to unify and generalize collection of distinct results - how to see DPG, actor-critic, variance reduction in stochastic nets, action-conditional baselines,synth grads.. all from common lens?

    Prikaži ovu nit
    Poništi
  16. 8. sij 2019.

    A must-read review of the literature on model-based RL, world models using DL and how it compares to analogues in human cognition (mental sim, imagination). Lots of work left, in particular regarding finding the right abstractions, deal with time, and building partial models!

    Poništi
  17. proslijedio/la je Tweet

    I hear the GOP thinks women dancing are scandalous. Wait till they find out Congresswomen dance too! 💃🏽 Have a great weekend everyone :)

    Poništi
  18. 21. pro 2018.

    The barber's paradox of papers (kind of).

    Poništi
  19. proslijedio/la je Tweet
    9. pro 2018.
    Odgovor korisnicima i sljedećem broju korisnika:

    My personal journey into Belief Propagation is narrated here , a chapter we had to discard from for space considerations. But many find it educational, especially the story about Bill Gates.

    Poništi
  20. proslijedio/la je Tweet
    8. pro 2018.

    Our poster "Deep Learning for Classical Japanese Literature" is here. Thank you

    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:

    ·