Christian Szegedy

@ChrSzegedy

, research scientist. Opinions are my own.

Vrijeme pridruživanja: lipanj 2015.

Tweetovi

Blokirali ste korisnika/cu @ChrSzegedy

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

  1. 28. sij
    Poništi
  2. 25. sij

    also neural networks. that's why hard negative mining works

    Poništi
  3. 25. sij

    nice collection of alternative generation orders of (text) sequences

    Poništi
  4. 23. sij

    Amazing application of computer vision on computer graphics.

    Poništi
  5. 23. sij

    A huge step forward for neural science. It's exciting to see how artificial neural networks and machine perception plays an enabling role in analyzing and understanding the working of biological neural systems.

    Poništi
  6. 14. sij

    Mathematical Reasoning in Latent Space will be featured at . Multiple step reasoning can be performed on embedding vectors of mathematical formulas.

    Poništi
  7. 14. sij

    A really nice Jupyter notebook demonstrating the use of the API of the HOList theorem proving environment:

    Poništi
  8. 2. sij
    Poništi
  9. 17. pro 2019.

    Another (simplest) case of more data harming in machine learning: well specified, linear MSE, via gradient descent.

    Poništi
  10. 16. pro 2019.
    Poništi
  11. 12. pro 2019.

    Interested in AI for higher order logic theorem proving? Interfacing with DeepHOL/HOL-Light has been made simpler, demo code and jupyter notebook: Great work by Jordan Starkey, Tyler McFadden, Aaron Hadley and Sumit Jha

    Poništi
  12. 30. stu 2019.

    New efficient normalization for convolutional networks.

    Poništi
  13. 13. stu 2019.
    Poništi
  14. 12. stu 2019.

    A cool semi-supervised training trick.

    Poništi
  15. 12. stu 2019.

    Graph neural networks for higher order theorem proving will be featured at AAAI 2020.

    Poništi
  16. 2. stu 2019.
    Poništi
  17. 30. lis 2019.

    New open-access journal on combinatorics

    Poništi
  18. 17. lis 2019.
    Poništi
  19. 16. lis 2019.

    A really cool study which makes perfect sense, but nice to see it evaluated.

    Poništi
  20. 10. lis 2019.

    Maybe that's what quantum mechanics is about as well: the simulator predicts a branched future and then collapses it according to our choices. ;)

    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:

    ·