Matthias Bethge

@MatthiasBethge

Computational Neuroscience, Machine Learning, Deep Neural Networks

Tuebingen
Vrijeme pridruživanja: ožujak 2012.

Tweetovi

Blokirali ste korisnika/cu @MatthiasBethge

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

  1. proslijedio/la je Tweet
    16. sij

    Our new adversarial attack published @ NeurIPS 2019 is now available for Foolbox and CleverHans! The attack is SOTA in L0, L1, L2 & Linf, needs close to no hyperparameter tuning & is less susceptible to some types of gradient masking. Blog post @

    Prikaži ovu nit
    Poništi
  2. proslijedio/la je Tweet
    15. stu 2019.

    Check out our work on modeling the mouse visual system with pretrained networks trained on static object recognition. We find that equivalent random networks are just as useful! Come by if you’ll be at . Colab of

    Prikaži ovu nit
    Poništi
  3. 12. stu 2019.

    Great example for the spirit of breaking new ground for research excellence in Europe: is planning an outside-the-box AI unit to attract top machine learning researchers to Alicante:

    Poništi
  4. 14. ruj 2019.

    Bernhard Schölkopf receives big award for his pioneering work and impact in machine learning & AI and his efforts to help shaping the future in Europe ().

    Poništi
  5. proslijedio/la je Tweet
    14. ruj 2019.
    Poništi
  6. proslijedio/la je Tweet
    7. srp 2019.

    We developed gradient-based versions (L0, L1, L2 & Linf) of our Boundary adversarial attack that (1) resist gradient-masking, (2) perform better & are more query-efficient than SOTA (e.g. PGD or C&W) and (3) require virtually no hyperparameter tuning:

    Prikaži ovu nit
    Poništi
  7. proslijedio/la je Tweet
    13. lip 2019.

    📣📣We are hiring! 📣📣 We are searching for a Lead Machine Learning Engineer. You will co-lead our -team and work together with our tech and business team to conceptualize, develop and refine use-cases, models and stand-alone products. Apply at:

    Poništi
  8. 18. svi 2019.

    Scary to watch politicians of democratic countries betraying democracy

    Poništi
  9. proslijedio/la je Tweet
    8. svi 2019.

    We are hiring! Layer7 AI is looking for (Lead) ML Engineers and Interns who like to leverage promising datasets to either identify, train & implement customized AI models for our clients or to build standalone scalable AI products. Check out our openings:

    Poništi
  10. proslijedio/la je Tweet
    25. ožu 2019.

    1/6 Deep classifiers seem to be extremely invariant to *task-relevant* changes. We can change the content of any ImageNet image, without changing model predictions over the 1000 classes at all. Blog post @ . with Rich Zemel

    Prikaži ovu nit
    Poništi
  11. 9. pro 2018.

    Spontaneous reunion dinner at in Montreal. I had a really nice time seeing my great coworkers again.

    Poništi
  12. 6. pro 2018.

    Machine Learning scientists founded the ELLIS society today at and in Montreal. Industry, science organizations and top researchers in the field are . Full story: and German press:

    Poništi
  13. proslijedio/la je Tweet
    24. stu 2018.

    2.0 is released! Python package, GUIs, new user guide, use GPUs if you don't have your own -> new examples, including cheetah hunting! 🐆Massive congrats to the team: , , !

    Prikaži ovu nit
    Poništi
  14. 9. stu 2018.

    Great results, looking forward to the workshop!

    Poništi
  15. 10. ruj 2018.

    Rocket business in the Cyber Valley...

    Poništi
  16. proslijedio/la je Tweet
    13. srp 2018.

    Panel discussion on the initiative w/

    Poništi
  17. proslijedio/la je Tweet
    11. srp 2018.

    Do you have an idea on how to attack neural network models using only a small number of queries? Or do you have an idea on how to defend against adversarial attacks? Join our NIPS 2018 Adversarial Vision Challenge! .

    Poništi
  18. 25. svi 2018.
    Poništi
  19. 21. tra 2018.

    Selective attention is ubiquitous in human vision but plays hardly any role in computer vision yet. "Cluttered omniglot"---a one-shot segmentation task in clutter---could be a minimalistic but surprisingly challenging benchmark that may reveal the usefulness of attention.

    Poništi
  20. 20. tra 2018.

    Is your network less easily fooled than others? Here is a great opportunity to show it...

    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:

    ·