Bhav Ashok

@bhavashok

Computer Vision & Machine Learning Researcher | Founder | Pioneer | Computer Vision | CS + Math |

Vrijeme pridruživanja: travanj 2014.

Tweetovi

Blokirali ste korisnika/cu @bhavashok

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

  1. proslijedio/la je Tweet
    3. velj

    I'm bullish on many of the things Bhav says, they're building some incredible things over at .

    Poništi
  2. 3. velj

    Though it's fashionable these days to claim that we're in an AI winter, deep learning still has many extremely valuable opportunities. In this post, I express some of my 'definite optimism' and share some of these opportunities.

    Poništi
  3. 27. sij

    In this post, I outline a potential solution to the Deep Learning reproducibility problem

    Poništi
  4. 16. sij

    Imagine being a kid and seeing something this magical. This is going to inspire a generation of really motivated AR creatives. I look forward to the future.

    Poništi
  5. 10. sij

    Suppose we have 4 people in a room, 2 colorblind and 2 not. One non-colorblind person leaves the room. Who's right? 2 colorblind people or the other? There's a single reality but 2 sides that disagree, surely one of them is wrong right?

    Poništi
  6. proslijedio/la je Tweet
    25. pro 2019.

    A new blog post on one of the most useful metrics used in deep learning, Precision-Recall

    Poništi
  7. proslijedio/la je Tweet
    30. pro 2019.

    One thing that is useful is if someone can think coherently: it is totally possible to think your way to novel ideas or solutions. There is an idea maze and you can describe the walls and constraints. The big problem is thinking coherently is surprisingly rare.

    Poništi
  8. proslijedio/la je Tweet
    25. stu 2019.

    Here's my conversation with Gilbert Strang, a professor of mathematics at MIT & an inspiring teacher of linear algebra to millions of students around the world through MIT OpenCourseWare. I was one of those students and am forever grateful for the journey.

    Poništi
  9. proslijedio/la je Tweet
    24. stu 2019.

    Check out this episode of Forward Thinking Founders, which is an intro to neural networks. Bhav was an engineer at and before founding . We go deep in this episode. Check it out!

    Poništi
  10. proslijedio/la je Tweet
    5. ruj 2019.

    If you'd like to hear the whole story as to how we got from the movie on the left to the movie on the right, come join me for a "tweetorial" starting now!

    Prikaži ovu nit
    Poništi
  11. proslijedio/la je Tweet
    3. lis 2019.

    Now for something different! Deep RL + GAN training + CelebA = artificial caricature. Agents learn to draw simplified (artistic?) portraits via trial and error. @ creativity workshop. Animated paper: PDF: Thread.

    Prikaži ovu nit
    Poništi
  12. proslijedio/la je Tweet
    7. kol 2019.

    In safety-critical situations like autonomous driving, neural networks aren't good enough at visual recognition—yet. shows how Nuron Labs can change that:

    Poništi
  13. 6. kol 2019.

    Architecture: Brain as Weights: Learned knowledge, the structure compensates for what the environment/supervision/optimization lacks

    Poništi
  14. proslijedio/la je Tweet
    17. srp 2019.

    "Building a worldwide neural layer"

    Prikaži ovu nit
    Poništi
  15. 14. srp 2019.

    I will be presenting my work on building a worldwide neural layer that's going to bring fast and reliable neural networks to the real world.

    Poništi
  16. proslijedio/la je Tweet
    1. srp 2019.

    Objects emerge in 3D-bottlenecked RNNs supervised by moving and watching objects move, in new work by Adam and team

    Poništi
  17. proslijedio/la je Tweet
    10. svi 2019.

    : a strange imagenet-like dataset with very wrong-looking labels, yet a model trained on it does totally well on the normal validation set. It's a crime against ML!

    Poništi
  18. proslijedio/la je Tweet

    Which came first, the chicken or the egg? At last, the definitive answer to this age-old question…

    Poništi
  19. proslijedio/la je Tweet
    20. pro 2018.

    imitating videos with BigGAN: first predict Y from the source, then feed it to the generator. input here is the trailer for Planet Earth

    Prikaži ovu nit
    Poništi
  20. proslijedio/la je Tweet
    1. velj 2019.

    This is a super cool resource: Papers With Code now includes 950+ ML tasks, 500+ evaluation tables (including SOTA results) and 8500+ papers with code. Probably the largest collection of NLP tasks I've seen including 140+ tasks and 100 datasets.

    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:

    ·