Wieland Brendel

@wielandbr

Machine learning researcher in the , co-founder of and lead organiser of the German school competition on AI ().

Vrijeme pridruživanja: kolovoz 2015.

Tweetovi

Blokirali ste korisnika/cu @wielandbr

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

  1. Prikvačeni tweet
    6. velj 2019.

    Neural networks seem to use a puzzlingly simple strategy to classify images (work accepted at ICLR 2019 and liked by ;-)). Digest @ 1/8

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

    Does your DNN have problems with common corruption robustness? You can get suprisingly far by just training on noise! In our new paper, we evaluate how simple learned i.i.d. noise can help to generalize to ImageNet-C. Blog post @

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

    Most expenses should be reimbursed no questions asked, with a random audit system to follow up.

    Poništi
  4. 16. sij
    Prikaži ovu nit
    Poništi
  5. 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
  6. proslijedio/la je Tweet
    10. pro 2019.

    Signature of the partnership between ELLIS and the CIFAR Learning in Machines and Brains program. With Nuria Oliver, Bernhard Schölkopf, Yoshua Bengio and the members of ELLIS at NeurIPS.

    Prikaži ovu nit
    Poništi
  7. 10. pro 2019.

    It's really amazing to see how far the ELLIS Initiative has come within one year!

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

    It was a great pleasure supporting the first school competition on AI in Germany. The quality of the teams was really impressive! We are looking forward to round two next year!

    Poništi
  9. proslijedio/la je Tweet
    20. stu 2019.

    Neuroscience has inspired , but lacks methods to directly translate neural data into better algorithms. Lead by Zhe Li in our paper we used neural data to engineer more robust AI algorithms with better generalization .

    Prikaži ovu nit
    Poništi
  10. proslijedio/la je Tweet
    4. ruj 2019.

    We are hiring!📣 Build exciting solutions with us & help shape a dynamic startup from its early days. If you know how to work with common deep learning frameworks & have an entrepreneurial mindset, join us as a Machine Learning Engineer.

    Poništi
  11. proslijedio/la je Tweet
    22. kol 2019.

    We're releasing a new method to test for model robustness against adversaries not seen during training, and open-sourcing a new metric, UAR (Unforeseen Attack Robustness), which measures how robust a model is to an unanticipated attack:

    Poništi
  12. proslijedio/la je Tweet
    18. srp 2019.

    Can your object detector handle noise? What will your autonomous vehicle do when it's foggy? And more importantly: Is your D(rago)NN ready for Snow when Winter is Coming? Find out with our robust detection benchmark (Pascal, Coco & Cityscapes):

    Poništi
  13. 7. srp 2019.

    Code and another extension of the paper will follow soon. Concurrent work by the group of M. Hein, who published just one day after us, has found similar results with almost the same technique:

    Prikaži ovu nit
    Poništi
  14. 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
  15. proslijedio/la je Tweet
    1. srp 2019.

    Article discussing our recent work investigating the properties of current state-of-the-art algorithms. Hope to see more in this line of research!

    Poništi
  16. proslijedio/la je Tweet

    We are happy to announce the start of the organized by , and . Participate now and bring disentangled representations to the REAL World - with 20k€ in prices & best paper awards!

    Poništi
  17. 13. lip 2019.

    Layer7 AI is deeply embedded into the world-class Tübingen ML landscape and is committed to employ state-of-the-art machine learning techniques to solve our clients problems. This is a great opportunity for your career!

    Poništi
  18. proslijedio/la je Tweet
    7. lip 2019.

    If you're in this afternoon and interested in what we do at , join us for free drinks & tarte flambee at 4:30pm in the common area of the Tübingen AI Research Building

    Poništi
  19. 8. svi 2019.
    Poništi
  20. proslijedio/la je Tweet
    30. tra 2019.

    three phd positions are available in the excellence cluster "machine learning - new perspectives for science" to work on (1) ML in linguistics and computer vision (2) ML in psychophysics and (3) ML in geoscience

    Poništi
  21. 18. velj 2019.

    We published guidelines for evaluating adversarial robustness (). This is a living document: please feel invited to join & share your thoughts. By Nicholas Carlini

    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:

    ·