Kevin Luxem

@kvnlxm

Doctoral researcher focusing on machine learning methods to understand how neural activity is linked to behavior, Physicist by training, Outdoors Enthusiast

Bonn, Germany
Vrijeme pridruživanja: siječanj 2018.

Tweetovi

Blokirali ste korisnika/cu @kvnlxm

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

  1. proslijedio/la je Tweet
    11. sij

    Having a research paper is a good way to demonstrate your skills for great school, but it's not everything, a thread:

    Prikaži ovu nit
    Poništi
  2. 11. sij

    Interesting work. It's still on my list to play with attention layer's in my sequence models. Hope I find some time soon 🤹

    Poništi
  3. 11. sij

    Just came across this highly interesting podcast on AI with . Why haven't I heard about it before? 🤔 It's so enjoyable to listen to it, especially the episode with

    Poništi
  4. 9. sij
    Poništi
  5. 9. sij

    This is fascinating and I'm eager to experiment with evolutionary algorithm. Kinda convinced that this will be the next step in ML 😎

    Poništi
  6. proslijedio/la je Tweet
    9. sij

    Congratulations to the new director of , !

    Poništi
  7. 5. sij

    From my own experience, also code obeys in some ways the laws of thermodynamics and can get nasty 😁 Improving it's readability and quality is key for reproducibility and correctness

    Poništi
  8. proslijedio/la je Tweet
    3. sij

    A General and Adaptive Robust Loss Function They propose an analytical function that can represent a family of well known robust cost functions just with a single parameter (alpha). Alpha lets you walk through L2, huber, cauchy, tukey and more.

    Prikaži ovu nit
    Poništi
  9. 3. sij

    In general, I support DeepMind, but this... Especially a journal like Nature should aim for quality and scientific discovery rather than looking at media coverage 🤔

    Tweet je nedostupan.
    Poništi
  10. 2. sij

    Reproducibility is a serious topic in any scientific endeavour and I'm excited to see the AI community is making a step towards it.

    Poništi
  11. 28. pro 2019.

    Nice overview of common regression losses in DL with their pro and cons. Definitely helped me to understand a nasty NaN problem with my model ...

    Poništi
  12. 26. pro 2019.

    "There are many results on arXiv that only work on a handful of datasets or work on bajillion GPUs that only Google infrastructure can support. Do the community a favor and stop publishing those as general results" It's general when it can cope with real world data, not MNIST 🎄

    Poništi
  13. proslijedio/la je Tweet
    26. ruj 2019.

    When you finish a PhD in computer science, they take you to a special room and explain that you must never use recursion in real life. Its only purpose is to make programming hard for undergrads.

    Poništi
  14. 23. ruj 2019.

    This is very exciting work and I enjoy cross-modality learning. I wonder to which applications this could lead. Definitely an interesting read!

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

    We've observed AIs discovering complex tool use while competing in a simple game of hide-and-seek. They develop a series of six distinct strategies and counterstrategies, ultimately using tools in the environment to break our simulated physics:

    Prikaži ovu nit
    Poništi
  16. 17. ruj 2019.

    Made it to and Starting with a promising satellite workshop.

    Poništi
  17. proslijedio/la je Tweet
    15. ruj 2019.
    Poništi
  18. 15. ruj 2019.

    Amazing talk by on advances in reinforcement learning . Very inspiring stuff Always love to see an agent playing Atari games 😂

    Poništi
  19. proslijedio/la je Tweet
    13. ruj 2019.

    Looking for a way to infer behavioral states in experiments combining population acitivity recordings and behavioral tracking via ? Have a look at our proceeding with and at or visit our conference poster at S1B56!

    Poništi
  20. 13. ruj 2019.

    Interested in deep behavior quantification and neuronal activity? Check out our poster tomorrow . Session PS1-B.55

    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:

    ·