Filip Piekniewski

@filippie509

I work on computer perception, mostly vision. Check the blog for details. Prev. & Opinions are obviously my own.

Vrijeme pridruživanja: listopad 2015.

Tweetovi

Blokirali ste korisnika/cu @filippie509

Jeste li sigurni da želite vidjeti te tweetove? Time nećete deblokirati korisnika/cu @filippie509

  1. Prikvačeni tweet
    24. pro 2016.

    Stacked predictive encoder with feedback. Simple idea with some surprising consequences:

    Poništi
  2. prije 9 sati

    On this day, 10 years ago I came on a plane and flew into the USA to start a new life adventure, equipped with 2 bags, one with clothes & one with some basic necessities (below). It's been a blast so far, I've met great people and seen amazing things and the adventure continues!

    Poništi
  3. This illustrates a general concept popular among some: that AI could be hacked together on a party by a bunch of tech bros. I'm sure they can hack together a quick demo leveraging spurious correlations, but a safe and reliable AI is not going to be a result of such approach.

    Poništi
  4. proslijedio/la je Tweet
    prije 24 sata
    Odgovor korisnicima

    You and your team will give up using statistical reasoning for real world decision making when you have a deep understanding on dynamical systems. I could come if invited but I surely would ruin the party. ;)

    Poništi
  5. 2. velj

    I thought they no longer have backup drivers in Arizona? Or perhaps it was just a publicity stunt...

    Poništi
  6. 1. velj

    Amazing how the seemingly smart social media companies don't realize the most obvious fact in the universe: if you ban something relatively high profile, it only becomes more popular and instantly goes viral.

    Poništi
  7. 31. sij

    Things you hear before a bubble blows up.

    Poništi
  8. proslijedio/la je Tweet
    31. sij
    Poništi
  9. 30. sij

    ROTFL. He really said this? There will be exactly zero Tesla robotaxis by the end of this year and likely same in 2021, 2022 and 2023. I once more reiterate my prediction.

    Poništi
  10. 29. sij

    Excellent episode of showing the connection between the period doubling bifurcation diagram and the Mandelbrot set, some pretty cool visualizations there. I think there is something fundamental there that we don't understand at all.

    Poništi
  11. proslijedio/la je Tweet
    28. sij
    Odgovor korisnicima

    Moravec's paradox in a nutshell: we still can't get machines to do anything nearly as complex as actually intelligent creatures, so let's make them play board games again instead, because in the eyes of the public (especially investors), "board game = smart"

    Poništi
  12. 28. sij

    If anything this indicates how relatively simple games are compared to anything of any significance in robotics i.e. the real world. Just yet another incarnation of Moravec's Paradox.

    Poništi
  13. proslijedio/la je Tweet
    25. sij

    Excellent and perceptive article by on the state of AI and whether machines could ever attain human-level intelligence or consciousness. Features discussion of recent books by & Ernest Davis; Christof Koch; and me.

    Poništi
  14. proslijedio/la je Tweet
    21. sij

    How long will those idiots parade this zombie puppet stupidity? Seriously, this is worse than 18th century mechanical turk.

    Poništi
  15. 22. sij

    LOL I clicked on the website to see when can I get one, there is a big apply button on the top so I'm thinking fine, maybe you need to apply to get one, and then it redirects me to cruise hiring website. Hilarious.

    Prikaži ovu nit
    Poništi
  16. 22. sij

    This is likely the peak of the self driving bubble. FYI: you won't see any of these and if you do they will be remote controlled.

    Prikaži ovu nit
    Poništi
  17. 21. sij

    How long will those idiots parade this zombie puppet stupidity? Seriously, this is worse than 18th century mechanical turk.

    Poništi
  18. proslijedio/la je Tweet
    19. sij
    Poništi
  19. 17. sij

    Excellent article describing the concept of a complex system and why we have conceptual problems describing them: I'm writing an article placing the problem of AI as optimization within a complex system, so this is a good pre-read.

    Poništi
  20. 17. sij

    As much as I agree that disengagement measurement for AV's can be manipulated and is largely meaningless, I wonder how Cruise plans to do statistical significance testing on an inherently long tail distribution.

    Poništi
  21. 17. sij

    Finally watched the Bengio Marcus debate. I've got somewhat mixed feelings. I feel it was somewhat vacuous and both used the same words to mean something else all the time and kinda talked over their heads...

    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:

    ·