Jörg Doku

@JorgDoku

Deep Learning, software engineering, and enterprise software. Ex- , , , . Marathon Runner. Deadlifts.

Vrijeme pridruživanja: studeni 2011.

Tweetovi

Blokirali ste korisnika/cu @JorgDoku

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

  1. proslijedio/la je Tweet

    If you liked shorting Tesla at $100, $200, $300, $400, $500 and $600, you're going to love shorting it at $700.

    Poništi
  2. proslijedio/la je Tweet
    30. sij

    3 things that help habits stick: 1) Repetition. Habits form based on frequency, not time. 2) Stable context. If the context is always changing, so is the behavior. You need a reliable environment. 3) Positive emotions. If it feels good, you’ll want to repeat it.

    Poništi
  3. proslijedio/la je Tweet
    24. sij

    "It's easier to hold your principles 100 percent of the time than it is to hold them 98 percent of the time." -- Clayton Christensen, giant of enterprise, RIP

    Poništi
  4. proslijedio/la je Tweet
    23. sij
    Poništi
  5. proslijedio/la je Tweet
    22. sij

    "Let me get this straight" is a signal that the opposite is about to happen.

    Poništi
  6. proslijedio/la je Tweet
    21. sij

    Apparently in Epic's bathrooms and all around the buildings, there's a list of 10 commandments. The first two are "Do not go public" and "Do not be acquired." (h/t )

    Prikaži ovu nit
    Poništi
  7. proslijedio/la je Tweet
    14. sij

    Here's my conversation with Daniel Kahneman, Nobel Prize winning economist and psychologist, author of Thinking Fast and Slow. We talk about two systems of thought in the human mind and their connection to deep learning, AI, and autonomous vehicles.

    Poništi
  8. proslijedio/la je Tweet
    13. sij

    I'm thrilled to finally tell the world about computer-designed organisms aka "xenobots" aka "living robots" aka "reconfigurable organisms"... You've heard of sim2real -- this is sim2life.

    Prikaži ovu nit
    Poništi
  9. proslijedio/la je Tweet
    14. sij

    When I flew out to sf over 5 years ago for my final interview at Plaid, almost all the “experts” I talked to told me not to join. “The market is too small, the founders don’t have the right experience, the team is all randos, the company is way too small, etc etc”

    Prikaži ovu nit
    Poništi
  10. proslijedio/la je Tweet
    13. sij

    I've been at this startup for less than a month and I've been exposed to so many problems I didn't even know existed. I'm of the increasing belief that everyone should try a startup early in their career, e.g. within the first 3 years/before settling into complacency.

    Poništi
  11. 12. sij

    If wisdom is knowing the long term consequences of your actions, then planning essentially makes you wiser.

    Poništi
  12. proslijedio/la je Tweet
    9. stu 2019.
    Odgovor korisniku/ci

    The No Free Lunch theorem is often misinterpreted. There are several universal approximation bounds (Cybenko, Barron, Vapnik) for ML model classes. These show that any reasonably behaved function can be learned, but the price of lunch is huge amounts of data

    Poništi
  13. proslijedio/la je Tweet
    Odgovor korisnicima

    Studies of the source of speedup across many areas has shown that roughly half of the improvement is due to hardware and half due to software. In some cases, algorithm advances have been much more important than hardware.

    Poništi
  14. proslijedio/la je Tweet
    23. kol 2019.

    Most startup founders end up spending their entire careers trying to catch lightning in a bottle even once. I caught it once and lost it myself. It’s hard to hold— so much can go wrong. When you catch lightning in a bottle, you have to do everything in your power to harness it.

    Prikaži ovu nit
    Poništi
  15. proslijedio/la je Tweet
    20. stu 2019.

    the LA consumer startup ecosystem is coming into its own -- Honey, Snap, Riot Games, Tinder, Bird, Dollar Shave Club. The most $1B+ consumer startups outside of the Bay Area? A few years back, I might have guessed that NYC would be the emerging leader. But pretty clear it's LA.

    Poništi
  16. 10. sij

    We'll have to rely on other methods of ensuring trust, e.g. cryptography.

    Prikaži ovu nit
    Poništi
  17. 10. sij

    I think deep fake detection software is a plausible solution, but it'll require tons of computation, which over time is becoming more expensive. Such solutions will likely be limited to large corporations, whose best interest profit-wise might not be the detection of deepfakes.

    Prikaži ovu nit
    Poništi
  18. 10. sij

    I think fake audio / video won't easily be detected in the future, and we'll learn to stop trusting content using natural approaches (recognizing the person's face and/or voice).

    Prikaži ovu nit
    Poništi
  19. 10. sij

    There's going to be an arms race between white hat AI groups and black hat AI groups. I just don’t see the white hats keeping the pace with the people who are trying to fool everyone.

    Prikaži ovu nit
    Poništi
  20. proslijedio/la je Tweet
    4. lis 2019.

    It is irresponsible and selfish to be teaching grad students and postdocs MPI instead of Spark or Tensorflow or Dask or almost literally anything else.

    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:

    ·