Karl Higley

@karlhigley

Recommender systems machine learning engineer. Privacy-preserving ML . Autistic. He/him. Prev: , , ,

Brooklyn, NY
Vrijeme pridruživanja: siječanj 2009.

Tweetovi

Blokirali ste korisnika/cu @karlhigley

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

  1. Prikvačeni tweet
    10. lip 2019.

    Some thoughts on recommender systems, a thread of threads. 🧵

    Prikaži ovu nit
    Poništi
  2. proslijedio/la je Tweet

    This is cyberpunk. Changing the physical changes the digital which changes the physical. Power held by governments and corporate powers can be subverted and redirected by regular people who understand how the system feeds upon itself.

    Poništi
  3. proslijedio/la je Tweet
    2. velj
    Poništi
  4. proslijedio/la je Tweet

    In neurotypical society, "empathy" is often confused for "sympathy". I.e., if you understood, you'd do the same as me. But even when we do understand, NDs are often still able to see all the possible alternative choices.

    Poništi
  5. proslijedio/la je Tweet
    prije 18 sati

    the downstream impact of autocratic leadership is almost as fascinating as it is horrifying

    Poništi
  6. prije 19 sati

    As an autistic person, I have experienced this in almost every work environment. To me, it seems very weird. Aren’t we supposed to care about our coworkers (and ourselves)? Why is being pleasant more important than being kind?

    Poništi
  7. proslijedio/la je Tweet
    2. velj
    Odgovor korisniku/ci

    They strongly believe only autistics are obsessed by routine and ritual, but their own society is full of rituals they no longer perceive, such as the "How are you?" ritual and the "on trend" obsession of clothing style conformity.

    Poništi
  8. proslijedio/la je Tweet
    prije 19 sati
    Poništi
  9. proslijedio/la je Tweet
    2. velj

    "It depends" sounds too wishy washy and is not a great conversation starter/continuer. For the same semantics but a much better outcome, just ask "When?" There's almost always a when-to and when-not-to. Everyone learns more from the when discussion.

    Poništi
  10. proslijedio/la je Tweet
    prije 22 sata

    This is an excellent example of finding what is being measured and figuring out how to impact those variables to produce a desired outcome. Hacking 101

    Poništi
  11. proslijedio/la je Tweet
    2. velj

    The problem of wrapping a cool method up into a tool may seem like a detail or an afterthought in the development process, but in fact it deserves a pretty careful eye.

    Prikaži ovu nit
    Poništi
  12. proslijedio/la je Tweet
    2. velj

    This also comes up when I discuss software engineering roles versus research roles at Google. People finishing their PhDs are often stuck on only valuing the research roles, even though product teams have deep, interesting problems—and tackling them can create huge impact!

    Prikaži ovu nit
    Poništi
  13. 2. velj

    Some potentially interesting models for recommender system applications (where tabular data is common.)

    Poništi
  14. proslijedio/la je Tweet
    29. sij

    Tensor / array library developers! Please save the date for the Tensor Developer Summit, March 19–20, . Registration opens soon.

    Poništi
  15. proslijedio/la je Tweet
    1. velj

    Every medical-related GoFundMe campaign is a tragic policy failure

    Poništi
  16. proslijedio/la je Tweet
    24. sij

    Nobody likes big software abstractions, until we've written tons of code to work around not having those abstraction layers in the first place. E.g. : data repository pattern vs. direct database access. Integrate patterns into your skill set, to readily anticipate the future.

    Poništi
  17. proslijedio/la je Tweet
    31. sij

    "You don’t want to hear that the reason your monolith is a spaghetti monster is because you let it become that way, one commit at the time, due to weak habits, pressurized deadlines, or simply sheer lack of competence."

    Poništi
  18. proslijedio/la je Tweet
    31. sij

    What are some reasons we can’t forecast properly? 1. Start date uncertainty 2. The assumption that “size” of work is the main contributor to lead-time (when it's idleness) 3. Over-utilization – people not working on what you think they are, but totally busy 4. Dependencies...

    Poništi
  19. proslijedio/la je Tweet

    Your organization will not acquire machine learning fluency with one big breakthrough project. You will build the right processes, mindset, and tooling over time through a series of small successes across a diverse range of problems.

    Poništi
  20. proslijedio/la je Tweet
    31. sij

    I read a line in Rebels at Work (an excellent guide book, defs recommend) that said, “organizations want the appearance of diversity but not their impact.” to extend that: I also think organizations want the *appearance* of change but not its impact

    Prikaži ovu nit
    Poništi
  21. proslijedio/la je Tweet
    31. sij
    Odgovor korisniku/ci

    Me, autistic, for the next hour or two after the exchange: Did I mess up this interaction I really think I did but I don't know did I mess up this interaction I really think I did but I don't know did I mess up this interaction I really think I did but I don't know did I mess up

    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:

    ·