DeWitt Clinton

@dewitt

Research and Machine Intelligence at Google. Father, husband, and Eph. Mostly. (he/him)

Seattle, WA
Vrijeme pridruživanja: siječanj 2007.

Tweetovi

Blokirali ste korisnika/cu @dewitt

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

  1. Prikvačeni tweet
    16. svi 2019.

    1/ Some personal news: After almost 13 years working on what we now know as Google Cloud, I'm making a huge leap and taking on a new role at Google.

    Prikaži ovu nit
    Poništi
  2. prije 13 sati

    Does anyone know someone with root@nvidia? I'm in a back-and-forth discussion with a customer service rep trying to change my email address on my Nvidia account. They're well-meaning enough, but not empowered enough to fix it.

    Poništi
  3. 28. sij

    I've been using computers for going on 40 years and I still have no idea how to type a "—" or a "…" on my machine. I Google "unicode emdash" and "unicode ellipsis" and cut-and-paste. Every time. And I use them both daily. What's your dumbest technical confession?

    Poništi
  4. 28. sij

    To appreciate how non-discoverable multitasking is on the iPad, consider that I've used an iPad for 10 years and I only just now learned how to open a second app side-by-side via 's article here: Aaaand I've already forgotten it again.

    Prikaži ovu nit
    Poništi
  5. 28. sij

    A fundamental flaw of recommender systems is whatever braindead driveI I watch late at night on YT before bed (itself because of pretty good algorithmic recommendations!) does not imply those are topics I want pushed to me all day in discovery feeds. Feedback loops gone awry!

    Prikaži ovu nit
    Poništi
  6. proslijedio/la je Tweet
    21. sij

    Ironically if you cut Infinite Jest in half you end up with two Infinite Jests

    Poništi
  7. 12. pro 2019.

    Very excited to see "Advances and Open Problems in Federated Learning", a collaboration among many top researchers in the field, curated by our teammates Peter Kairouz and Brendan McMahan of Google. I expect these seeds will bear fruit for years to come.

    Poništi
  8. 11. pro 2019.

    My teammate speaks at TensorFlow World about his work on TensorFlow Federated: A great introduction to the emerging field of privacy-preserving machine intelligence.

    Poništi
  9. 30. stu 2019.

    I disabled all of my ad targeting preferences except one. I'm excited to see what full-on "Sandwich Twitter" is like.

    Poništi
  10. 16. svi 2019.

    15/ Phew, that was long! This will ideally be the last I tweet about myself for at least another 13 years. I hope to serve you all well in my new role. See everybody in the future!

    Prikaži ovu nit
    Poništi
  11. 16. svi 2019.

    14/ Where I'm headed reminds me in so many ways of the Google I joined so many years ago, full of brilliant people, tremendous kinetic energy, and seemingly impossible challenges to overcome. It's science fiction turned reality, and I'm a kid in a candy store once again.

    Prikaži ovu nit
    Poništi
  12. 16. svi 2019.

    13/ I feel like a Noogler all over again (Google's pet term for new hires), and I'm in equal measures terrified and giddy to dive head first into a new part of the company and into a whole new science.

    Prikaži ovu nit
    Poništi
  13. 16. svi 2019.

    12/ The team I'm joining in Seattle is at the intersection of cutting-edge applied research, massive real-world customers, and production Google-scale infrastructure, with a healthy dose of open source as well!

    Prikaži ovu nit
    Poništi
  14. 16. svi 2019.

    11/ And 's awesome recent oped in the NYT, "Privacy Should Not Be a Luxury Good", which touches on how we're already using federated learning to protect millions of people's personal privacy today:

    Prikaži ovu nit
    Poništi
  15. 16. svi 2019.

    10/ Or if visuals are more your style, this primer does an amazing job telling the story in comic-book form:

    Prikaži ovu nit
    Poništi
  16. 16. svi 2019.

    9/ As a high-level introduction, see Brendan McMahan and Daniel Ramage's post, Federated Learning: Collaborative Machine Learning without Centralized Training Data:

    Prikaži ovu nit
    Poništi
  17. 16. svi 2019.

    8/ For those unfamiliar with the field, Federated Learning is a branch of machine intelligence that explores what is possible when we distribute machine learning to the edge while preserving privacy and protecting personal data.

    Prikaži ovu nit
    Poništi
  18. 16. svi 2019.

    7/ They must have liked me, because starting in a week I'm joining the Federated Learning team here in Seattle. (I once joked that when the AI team at Google offers you a job, you say yes. Turns out, you do!)

    Prikaži ovu nit
    Poništi
  19. 16. svi 2019.

    6/ I explored everything from other teams, to other companies, to starting my own, but there was one opportunity that I knew from the very first conversation that I'd accept on the spot if they offered me the job.

    Prikaži ovu nit
    Poništi
  20. 16. svi 2019.

    5/ That said, I'm already 20 years deep into this crazy career, and people like me grow best by running headlong into the unknown, so it was time for me to disembark and try my hand at something new.

    Prikaži ovu nit
    Poništi
  21. 16. svi 2019.

    4/ The relationships I made, the customers we served, and the products we shipped all mean so much to me, and I want to hang on to those memories and friendships forever. I've been so lucky to have part of something so much bigger than I ever could have imagined.

    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:

    ·