Bindu Reddy   

@bindureddy

AI/ML research start-up founder & CEO, ex-GM and creator of AI verticals at AWS, and ex-Googler. Passionate about artificial intelligence and human behavior

San Francisco, CA
Vrijeme pridruživanja: svibanj 2007.

Tweetovi

Blokirali ste korisnika/cu @bindureddy

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

  1. Nowadays, Superbowl and other big games are pretty much the only thing that brings the country together... Good thing we have these games to celebrate and unify around. While we I am supporting the 49ers, here is hoping for a fast-paced good game

    Poništi
  2. 1. velj

    Coronavirus and other epidemics can be controlled by modeling and predicting their spread using initial reports, transmission mechanisms, and passenger movements. The Chinese gov't didn't choose to use to provide warnings and tried to be secretive & cover up the early cases! :(

    Poništi
  3. 31. sij

    The best way to learn is by actually doing things! Our first developer workshop and the first look at our deep learning APIs

    Poništi
  4. 30. sij

    I am pleasantly surprised by the rise of PyTorch over the last few years. A couple of years ago, it seemed like TensorFlow was the only game in town. Thanks to the superior debugging and prototyping capability PyTorch seems like its making some big in-roads :)

    Poništi
  5. 30. sij

    Based on earnings to-date, a simple deep-learning model would predict a very strong season for earnings and Dow potentially hitting 30K this quarter :) Tesla, Microsoft, and Apple all have had great earnings!

    Poništi
  6. 29. sij

    Experience the Magic of Generative Adversarial Networks (GANs)! An under the hood blog post about how we use GANs and the research we have done thus far -

    Poništi
  7. 29. sij

    Techcrunch on our launch. The post talks about deep reinforcement learning and generative models. We expect to see multiple deep-RL systems in production this year.

    Poništi
  8. proslijedio/la je Tweet
    28. sij

    Few products/entrepreneurs leave you excited enough in the first 30 mins to want to invest in and support the team. and is definitely one of them. Excited for their launch today...

    Prikaži ovu nit
    Poništi
  9. 28. sij

    We are out of stealth and are packaging generative models and neural architecture search into a shiny new AI service!

    Poništi
  10. 27. sij

    Thanks, Google for Dataset Search, now how about some much-needed improvements on Scholar? :) - Sort by number of citations - Search inside the article - Tag papers with code, datasets that they reference. Allow paper authors to edit/correct their entries.

    Poništi
  11. 26. sij

    More often than not, there are close to infinite possibilities... For example, anytime you shuffle a card deck, you are more likely than not holding an arrangement that has never been seen before and will never be seen again.

    Poništi
  12. 25. sij

    London police just announced that they are going to use facial recognition for mass surveillance. This is very unwise and could lead to the end of anonymity, free protests and take away some basic civil liberties. A case of good intentions can have bad unforeseen consequences

    Poništi
  13. 24. sij

    Google Dataset Search is cool, but the real problem is that there aren't enough robust datasets available in the public domain except for the few on Kaggle/UCI. It would be great if companies published interesting datasets instead of open-sourcing yet another ML library.

    Poništi
  14. 24. sij

    What deep learning practitioners do from different points of view. I especially appreciated the mathematician's point of view

    Poništi
  15. 22. sij

    Data science and good ML modeling applied to content marketing, Facebook and TV ad buying will be the deciding factor in the election. The campaigns who are savvy enough to judge and hire good technical talent will out-execute mediocre operations

    Poništi
  16. 21. sij

    It's a good thing that double-blind reviews are the norm in all the top machine learning conferences. This means even self-taught independent researchers can publish in top conferences and there is no 'reputable author' selection bias.

    Poništi
  17. 20. sij

    Facebook's work on using neural networks (NNs) to solve math equations is pretty imaginative. They convert math equations to a language-like syntax, use seq2seq and train on a huge dataset. Given enough examples, the NN can learn anything :)

    Poništi
  18. 19. sij

    If I were to pick a Ph.D. topic for my thesis today, it would be around teaching AI models how to learn - i.e. learning to learn. Our brains evolved to learn how to learn as opposed to simply memorizing specific tasks like walking. That's fundamentally what makes us so adaptable

    Poništi
  19. 18. sij

    Over thousands of years, democracy and capitalism have evolved to be the most successful systems because we have constantly tweaked these systems to make them better We will need to continue to tweak them to fix climate change, income inequality and foster fair competition

    Poništi
  20. 17. sij

    Google's Reformer paper is a much bigger deal than it is being made out to be. BERT and GPT-2 were not practical given the insane training times/costs. Reformer has the potential to cause a mini NLP revolution over the next few months! -

    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:

    ·