Sam Sinai

@samsinai

Co-founder and ML scientist . Interested in AI, fitness landscapes, viruses, and the origins of life. Iranian-American. + alum.

Cambridge,MA
Vrijeme pridruživanja: srpanj 2010.

Tweetovi

Blokirali ste korisnika/cu @samsinai

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

  1. proslijedio/la je Tweet
    30. sij

    40x faster predictions for even the deepest random forests with FIL’s new sparse forest support -

    Poništi
  2. proslijedio/la je Tweet
    26. sij

    OK, been on a bit of a negative science tear, so here's a study that is both novel and cool IMO. Ever wondered why sperm/testis express ALL THE GENES? Well, it might be to prevent DNA mutations through a process the authors call transcriptional scanning.

    Prikaži ovu nit
    Poništi
  3. proslijedio/la je Tweet
    27. sij
    Prikaži ovu nit
    Poništi
  4. proslijedio/la je Tweet
    26. sij

    I know a foreign-born AI professor who can't travel outside the US because she might not be allowed back in. A co-worker of mine made a trip to Canada and was denied re-entry for a year. If we keep turning away talented people, the US will lose its status as a leader in research

    Poništi
  5. proslijedio/la je Tweet
    25. sij

    A concern beyond time to publication is cost. Is it worth spending $50000-100000 to get that publication in journal X? We should use $ to move science forward as much as possible—not always by answering high-bar reviews that fail to yield new insight.

    Poništi
  6. proslijedio/la je Tweet
    26. sij
    Poništi
  7. proslijedio/la je Tweet
    25. sij

    This Japanese robot sumo literally requires an infinite peak of responsiveness

    Poništi
  8. proslijedio/la je Tweet
    25. sij

    The Travelling salesman problem is one of the most well know NP-hard problem. Concorde’s solver can be used to solve exactly or approximately even large instances.

    Poništi
  9. proslijedio/la je Tweet
    25. sij

    I’m gonna wait until two anonymous bat virus genome experts sign off on this before I read it.

    Poništi
  10. proslijedio/la je Tweet
    24. sij

    This has been a really fun one! challenging the wisdom that one needs a lot of data to do ML guided biodesign. In fact, if you have to choose, think about getting HQ small-N data aligned w your endpoint, vs HT proxy data. low-N ML now helps you succeed w the former!

    Prikaži ovu nit
    Poništi
  11. proslijedio/la je Tweet
    23. sij

    5-year predictions 1. Active learning in biology - confluence of design (ML), build (large-scale dna synth/edit), and test (multiplexed assays) allows massive scale for exploring new function. Will require programmatic ways to explore that space. Expmnts no longer limiting. [1/n]

    Prikaži ovu nit
    Poništi
  12. proslijedio/la je Tweet

    I have a favor to ask. For 20+ years I've been working on a dream: to make all science funded by US taxpayers freely available to all. We are on the verge of achieving this. But we need to show that people care. So please, if you can, sign this letter:

    Poništi
  13. proslijedio/la je Tweet
    23. sij

    New & report has recommendations on preventing global biological catastrophes. Our conclusion: Now is the time to establish a more global approach to prevent deliberate or accidental misuse of DNA synthesis technologies. Read the report

    Prikaži ovu nit
    Poništi
  14. proslijedio/la je Tweet
    23. sij

    Here is the promised blogpost: on the motivation behind our modeling / ML approach for protein-peptide interactions. We'll likely have another post soon focused more on the (structural) biology.

    Prikaži ovu nit
    Poništi
  15. proslijedio/la je Tweet

    Wyss researchers have created a high-throughput platform to generate an Adeno-associated virus 2 (AAV2) library containing 200,000 variants, each carrying a distinct mutation in the virus capsid protein.

    Poništi
  16. proslijedio/la je Tweet
    21. sij

    may get yelled at for bringing this up but the nastiest move northeastern admin made during our grad student union drive was implying that our many iranian members’ info would be given to the trump admin if we unionized (via the NLRB)

    Prikaži ovu nit
    Poništi
  17. proslijedio/la je Tweet
    20. sij

    You’re not seeing what you think you’re seeing...just FYI. EVOLUTION IS NUTS. That is the tip of the viper’s tail. It’s the lure of the Iranian spider-tailed viper!!!

    Poništi
  18. proslijedio/la je Tweet
    21. sij

    Why hard-code your acquisition function when you can use reinforcement learning to train a NN to do it? Michael Volpp, Lukas Fröhlich, Kirsten Fischer, Andreas Doerr, Stefan Falkner, , Christian Daniel

    Poništi
  19. proslijedio/la je Tweet
    20. sij

    talking to about transformative research in his lab and startup spinouts like Dyno for

    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:

    ·