Rajesh S (aiexplorations)

@aiexplorations

Senior industry professional in the data and AI space. Interests include product/solution innovation, entrepreneurship, product strategy and problem solving.

Bangalore, India
Vrijeme pridruživanja: kolovoz 2011.

Tweetovi

Blokirali ste korisnika/cu @aiexplorations

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

  1. Prikvačeni tweet
    Poništi
  2. Natural systems often exhibit self-similar behaviour to an extent that the structure and function derive from the self-similarity. This isn't necessarily the case with NNs, since they're subject to data and constraints of behaviour (units, activ. funcs., hyperparams, etc)

    Poništi
  3. Makes me wonder about how computer vision methods (especially ML for computer vision) would process something like this. It is not just plausible that algorithms can be hacked but what would be interesting is to see if they too make similarly wrong inferences about images.

    Poništi
  4. These are ultimately similarity measures to an empirical model, which perhaps makes modeling social networks and contact networks "modelable". Perhaps such studies are being done in evaluating the extent of spread of (and precautions required for) epidemics?

    Prikaži ovu nit
    Poništi
  5. Another interesting possibility opens up when we consider that social networks (both strong and weak connections) are used to transmit viruses. For instance, we can build models of social propensity of individuals based on shared infrastructure, shared family, shared genes...

    Prikaži ovu nit
    Poništi
  6. As someone who is interested in the dynamics of systems, I'm finding the spread of viruses fascinating now. That empirical, dynamical models can actually model the behaviour is particularly interesting.

    Prikaži ovu nit
    Poništi
  7. proslijedio/la je Tweet
    3. velj

    This is a nice example of Simpson's paradox: cancer mortality rates are decreasing in each age group, but increasing overall.

    Poništi
  8. Benign cancers are curable and even early stage malignant cancers can be treated. Early detection is key. Be prepared, be aware and be safe!

    Prikaži ovu nit
    Poništi
  9. It is clear that cancer isn't due to one assignable cause - there are factors encompassing one's lifestyle, food habits, exercise and activity levels, and genetics. That said, early detection can ensure that we can catch at least common cancers.

    Prikaži ovu nit
    Poništi
  10. I'm glad that there's a lot of promising work being done at the intersection of computer vision and cancer detection/prediction. As someone who has seen close relatives become a victim of the disease, I wish we see more progress.

    Prikaži ovu nit
    Poništi
  11. Profound and humorous at the same time.

    Poništi
  12. This is an incredible talk at the intersection of data, systems, and post-truth discourse in our constantly degrading global society - with Tristan Harris and Yuval Harari

    Poništi
  13. What big firms possess in process orientation, small firms possess in the ability to channel, direct and focus talent. There is a certain size where doing one or the other is hard to do - and you need to do both. This is the space where clusterfugs happen.

    Prikaži ovu nit
    Poništi
  14. The truth is that clusterfugs are common enough in all startups, whether you're at size 3 or size 30 or size 300 (to go only by the headcount). The underlying reason for this is the complexity of communication, which is endemic to processes.

    Prikaži ovu nit
    Poništi
  15. Memorable video: and Bob Sutton discussing the "clusterfug" that accompanies many scaling startups:

    Prikaži ovu nit
    Poništi
  16. Process orientation is seen as the enemy of innovation by many in small/startup firms but can become an extremely valuable quality when startups begin to scale.

    Prikaži ovu nit
    Poništi
  17. Rishad Premji sounds extremely pragmatic in this article. Mature thinking like this can be a lifesaver for startups that enterprises may be interested in partnering with or investing.

    Prikaži ovu nit
    Poništi
  18. Then there's the challenge of building models that have a very narrow focus, which in itself isn't a bad thing, but will promise more than it can deliver if we were to extend the solution to new applications in the same sub-domain.

    Prikaži ovu nit
    Poništi
  19. ... and it definitely doesn't ensure that there is an improvement in your condition, skill or attitude owing to the knowledge present in the videos, lectures or other things you've come to access or possess thanks to the course.

    Prikaži ovu nit
    Poništi
  20. Perhaps the online learning revolution has more in common with the revolution in large scale automation or outsourcing of customer service in businesses all over. You can access all the content you want but it doesn't guarantee that your learning needs will be met...

    Prikaži ovu nit
    Poništi
  21. The diminishing role of mentors and gurus who are experienced professionals or academicians, that carefully advise learners is a matter of concern. A few videos are not a good substitute for good quality, in person mentoring and interactions.

    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:

    ·