Ariel Herbert-Voss

@adversariel

Adversarial machine learning and security, occasionally also math and dumb memes. Research scientist / CS PhD / cofounder

Cambridge & SF
Vrijeme pridruživanja: rujan 2013.

Tweetovi

Blokirali ste korisnika/cu @adversariel

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

  1. Prikvačeni tweet
    26. svi 2019.

    Some news: I’m writing a book for titled “The Machine Learning Red Team Manual”. My aim is to provide a practical guide for anyone interested in adversarial ML and red teaming as it relates to in-production ML systems. A short thread on why this project matters:

    Prikaži ovu nit
    Poništi
  2. 25. sij
    Poništi
  3. proslijedio/la je Tweet
    15. sij

    If you missed my talk at on facial recognition, check it out here!

    Prikaži ovu nit
    Poništi
  4. proslijedio/la je Tweet
    11. sij

    tickets are on sale now! We're returning to 751 D Park in April 17-19 and we want you to be there. We're bringing more hands-on Village experiences and more world-class speakers. Join us! Cash at the door option still available.

    Poništi
  5. proslijedio/la je Tweet
    3. sij

    Looking for smart & creative people to think and write about the most important issue of this decade🤓! Apply to be a China Cyber Policy Fellow at Harvard's D/L January 15th! DM for info

    Poništi
  6. proslijedio/la je Tweet

    It's 2020, and I had one goal for 2019, publish a video cabinet of hacker curiosities. Along with capturing bits and bytes of hacker history, it's configured as a resource for teaching. We hope to double the entries in 2020. Check it out here:

    Prikaži ovu nit
    Poništi
  7. proslijedio/la je Tweet
    28. pro 2019.

    I’ll be talking more Offensive Machine Learning !

    Poništi
  8. 28. pro 2019.

    Since its recovery from the Antikythera shipwreck in 1900 people have been building replicas to understand the mechanism better. My personal favorite is the Lego version from 2010

    Prikaži ovu nit
    Poništi
  9. 28. pro 2019.

    Love me some ancient computer chunks

    Prikaži ovu nit
    Poništi
  10. 28. pro 2019.

    I love historical technological artifacts and analog computers and am pretty stoked to see the fragments of the Antikythera mechanism IRL

    Prikaži ovu nit
    Poništi
  11. proslijedio/la je Tweet

    I suspect this talk is gonna be good just from the title ---> "Don't Red Team AI Like a Chump" by (she also knows her sh*t so that's likely to make it really good)

    Poništi
  12. proslijedio/la je Tweet

    just noticed a bunch of talks are finally up! starting with watching kill it

    Prikaži ovu nit
    Poništi
  13. proslijedio/la je Tweet
    Poništi
  14. 8. lis 2019.
    Poništi
  15. proslijedio/la je Tweet

    Thrilled to see that our panel had such reach! Thank you again to panelists and moderator ! We look forward to sharing the video recording soon 🤗

    Poništi
  16. proslijedio/la je Tweet
    23. ruj 2019.

    What happens when you take a generative network (like Google Smart Compose) and start asking it to complete unusual sentences, such as: "My social security number is..." ? Great insights into the power of networks to memorize information - wanted or not!

    Poništi
  17. 17. ruj 2019.

    Playing with ImageNet Roulette and I can only get it to classify my face as "face" (play with it yourself: )

    Poništi
  18. proslijedio/la je Tweet

    If you're interested in seeing some concrete examples of how machine learning systems can be biased, often in upsetting ways, as well as the critical role training data and labels play in ML systems, check out and twitter search results for 'imagenet'.

    Prikaži ovu nit
    Poništi
  19. 15. ruj 2019.

    More of an Islay person generally but this is v good

    Prikaži ovu nit
    Poništi
  20. 15. ruj 2019.

    In the Highlands enjoying a brief respite from MIT and the Media Lab s i t u a t i o n

    Prikaži ovu nit
    Poništi
  21. proslijedio/la je Tweet
    5. ruj 2019.

    New report by & Brian Friedberg for highlights the practices media manipulation. It's fantastic - read it. And, I'll add, it's implications are even greater, taken what we know about how journalists use social media. 1/x

    Prikaži ovu nit
    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:

    ·