AI Village @ DEF CON

@aivillage_dc

We are a community of hackers, ML researchers, and security data scientists focused on the use and abuse of AI - DM to join our Slack

Las Vegas, NV
aivillage.org
Vrijeme pridruživanja: kolovoz 2017.
57 fotografija ili videozapisa Fotografije i videozapisi

Tweetovi

Blokirali ste korisnika/cu @aivillage_dc

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

  1. proslijedio/la je Tweet
    27. sij

    A lot has been said about the dangers of large-scale language model trained on the internet, like GPT-2. We fine-tuned a version of GPT-2 to avoid generating descriptions of non-normative behavior (killing, suicide, things inconsistent with social norms)

    14 replies 68 proslijeđenih tweetova 266 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  2. proslijedio/la je Tweet
    24. sij

    Cities across the United States: We're banning all municipal use of facial recognition. London: We're going to conduct "live facial recognition" with all of our cameras across the city.

    25 proslijeđenih tweetova 56 korisnika označava da im se sviđa
    Poništi
  3. proslijedio/la je Tweet
    16. sij

    ECC cheat sheet by (I keep this by my desk) and CryptoTool 1.4

    10 proslijeđenih tweetova 9 korisnika označava da im se sviđa
    Poništi
  4. proslijedio/la je Tweet
    16. sij

    adversarial images for non-computer vision

    The Macrocilix maia moth confuses predators with wing patterns that mimic two flies feasting on bird droppings. It even releases a pungent odor to drive home the deception.
    36 proslijeđenih tweetova 167 korisnika označava da im se sviđa
    Poništi
  5. proslijedio/la je Tweet
    15. sij

    Also Sophos Data Science alum talking at about multi-task learning for malware detection here:

    3 proslijeđena tweeta 6 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  6. proslijedio/la je Tweet
    15. sij

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

    1 reply 6 proslijeđenih tweetova 18 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  7. proslijedio/la je Tweet
    28. pro 2019.

    I’ll be talking more Offensive Machine Learning !

    12 proslijeđenih tweetova 34 korisnika označavaju da im se sviđa
    Poništi
  8. proslijedio/la je Tweet
    23. pro 2019.
    Odgovor korisniku/ci

    There have been several people introduced to AI + security thru the , but it's hard to tell that from just going to Defcon. These villages are focal points for the community, not training grounds.

    1 proslijeđeni tweet 4 korisnika označavaju da im se sviđa
    Poništi
  9. proslijedio/la je Tweet
    16. pro 2019.

    Update: Given recent efforts by local legislatures to outlaw facial recognition tech and the protests in Hong Kong & mask ban, we have chosen to focus on the privacy issues of facial recognition for 28. If you have art, anti recognition or recognition tech — reach out!

    51 proslijeđeni tweet 112 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  10. proslijedio/la je Tweet
    15. pro 2019.

    Please stop describing AI safety as "long term, technical" safety. Ethics, fairness and AI for Social Good problems are also very much playing the long game & in many cases involve confronting highly technical fundamental ML challenges, requiring theoretical innovation to solve.

    26 proslijeđenih tweetova 131 korisnik označava da mu se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  11. proslijedio/la je Tweet
    10. pro 2019.

    📢📢With , released a battle tested taxonomy of how ML systems fail, by attackers or inherent design, for engineers and policy makers Blog: Paper: With 1/

    7 replies 80 proslijeđenih tweetova 140 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  12. proslijedio/la je Tweet
    3. pro 2019.

    The human body is a device that can be reverse-engineered, de-anonymized, mapped, tagged, tracked, and controlled

    16 replies 173 proslijeđena tweeta 197 korisnika označava da im se sviđa
    Poništi
  13. proslijedio/la je Tweet
    2. pro 2019.

    I often get asked about datasets, so I wrote a post on some good places to look, and some notes about going beyond them. If I've missed anything, please tell me and I'll add that in too.

    15 proslijeđenih tweetova 28 korisnika označava da im se sviđa
    Poništi
  14. proslijedio/la je Tweet
    2. pro 2019.

    SO EXCITED to share this work - "On the Legal Compatibility of Fairness Definitions", to be presented as an oral at the HCML workshop We discuss the prevalent misuse of legal terms in ML fairness literature & the many problems that causes.

    13 replies 73 proslijeđena tweeta 229 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  15. proslijedio/la je Tweet
    2. pro 2019.

    "Adversarial Learning of Deepfakes in Accounting": adversarial examples for accounting are now a research reality HT

    5 proslijeđenih tweetova 8 korisnika označava da im se sviđa
    Poništi
  16. proslijedio/la je Tweet
    25. stu 2019.

    Today in Google AI stereotyping. A thread if you wanna be mad. As I am wont, I am reverse image searching rando Twitter accounts that follow me to see if they're real people because they're usually bots. "Steve Willson"? Probs not real. 1/6

    38 proslijeđenih tweetova 47 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  17. proslijedio/la je Tweet
    27. stu 2019.

    Attackers will know your systems and programs better than you, and they’ll use that knowledge to commit fraud, send spam, host malware. Effective offensive security can’t just be about exploiting technical gaps, but also about finding ways to abuse tools in creative ways.

    15 proslijeđenih tweetova 48 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  18. proslijedio/la je Tweet
    26. stu 2019.

    I'll always associate deepfakes with the propaganda moral panic, but deepfakes will arguably proliferate because of *advertising* Imagine ads with models who look like you and your friends — CTR rates will jump. The 4 examples below were auto-generated based on "archetype faces"

    47 replies 381 proslijeđeni tweet 1.386 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  19. proslijedio/la je Tweet
    24. stu 2019.

    Strongly recommend perusing this great, super approachable deck by 's on AI snake oil. Especially if you will be wandering anywhere near a security conference any time soon, and absolutely if you'll be venturing out among security vendors.

    Much of what’s being sold as "AI" today is snake oil. It does not and cannot work. In a talk at MIT yesterday, I described why this happening, how we can recognize flawed AI claims, and push back. Here are my annotated slides:
    Prikaži ovu nit
    3 proslijeđena tweeta 16 korisnika označava da im se sviđa
    Poništi
  20. proslijedio/la je Tweet
    4. stu 2019.

    Me exploring my data worried about the assumptions I am making:

    48 replies 1.043 proslijeđena tweeta 7.618 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi

@aivillage_dc još nije objavio nijedan Tweet.

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:

    ·