Tweetovi

Blokirali ste korisnika/cu @mishraka

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

  1. 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
  2. 29. sij

    Both proposed simple architecture, hiding downstream complexities Raytracing: project a ray into the world and follow it to the light source. Only billions of rays to follow. CNNs: only ever perform convolutions repeatedly, but need millions of labeled data and expensive GPUs

    Prikaži ovu nit
    Poništi
  3. 29. sij

    The success of raytracing in computer graphics and CNNs in computer vision share some similarities. Both solved complex modeling issues. Raytracing: shadows, reflection CNNs: No explicit modeling, just provide lots of labeled data for any task

    Prikaži ovu nit
    Poništi
  4. 29. sij

    Brownrigg's ad for state senate in California looks so much like Bloomberg's Ad for president. I wonder if that was intentional.

    Poništi
  5. 27. sij

    Life's so fragile. RIP Kobe Bryant

    Poništi
  6. 26. sij

    This is true for technical ideas as well. You see everyone arguing about a topic everyday, and you feel you have a good basic idea of what it is when it is simply an illusion.

    Poništi
  7. 25. sij

    A picture is worth a thousand actions.

    Poništi
  8. proslijedio/la je Tweet

    Donald Trump must be convicted and removed from office. Because he will always choose his own personal interest over our national interest. Because in America, right matters. Truth matters. If not, no Constitution can protect us. If not, we are lost.

    Poništi
  9. 23. sij

    What's the marginal utility of yet another online course on same machine learning topic? In this new world with duplicate/similar content, we need a reliable source to rank and sort them for quality.

    Poništi
  10. 23. sij

    Imagine a world where either clockwise or anti-clockwise order was universally used, and you didn't have to check for it and re-order the vertices anywhere.

    Poništi
  11. 21. sij

    If you ever wondered about the strength of the desire to belong to your tribe, observe republican senators during the impeachment trial. There's no other incentive that explains their permanent genuflection to a rather incoherent and undeserving man on the planet.

    Poništi
  12. 20. sij

    An advice, contrary to the popular opinion, to fellow builders: Build before you sell, not sell before you build. The advice might be opposite for a natural salesperson. We should all play to your strength.

    Poništi
  13. 19. sij

    A danger of following idols is when you disagree with them, you usually blame yourself. And that can be dispiriting. Be inspired by others but be your own idol.

    Poništi
  14. 16. sij

    Choosing what to label, how to label it and the cost of labeling are as onerous as trying to think of a good algorithm to get the answer directly without any data at all. In fact, as you understand the domain and context, you start to trade data for algorithmic constraints

    Poništi
  15. 16. sij

    Learning without a direct goal of solving a real problem is ephemeral. We only retain knowledge we acquire in a well-defined context.

    Poništi
  16. proslijedio/la je Tweet
    14. sij

    Oh, and if you're interested in giving this approach a try, we have a great dataset for you to test your ideas on: Image网 (pronounced "Imagewang). You can find the dataset here:

    Poništi
  17. 15. sij

    I had always wondered about the bright shiny lights on both sides of an interviewee on TV. Today, I learned it is a standard three-point cinematic lighting design. Key Light, Fill Light, Back Light.

    Poništi
  18. proslijedio/la je Tweet

    American elections should be decided by the American people, not by the Russian Government. Retweet if you agree!

    Poništi
  19. proslijedio/la je Tweet
    14. sij

    At medical imaging we often need to make a little data go a long way. One under-appreciated approach for this is self-supervised learning. It's almost magical! I've written a little overview to help you get started. Let me know if you try it :)

    Prikaži ovu nit
    Poništi
  20. 14. sij

    Augmented Reality will continue to bring computer graphics and computer vision closer together.

    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:

    ·