Amit Sharma

@amt_shrma

Curiously curious, amusingly mechanic. Researcher at Microsoft Research. Work on causal inference, machine learning and experimentation.

New York
Vrijeme pridruživanja: listopad 2010.

Tweetovi

Blokirali ste korisnika/cu @amt_shrma

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

  1. Prikvačeni tweet
    21. kol 2018.

    If you want to learn more about applying causal inference, try out this Python library and I built. We've implemented many of the popular methods so that you can focus on the harder causal questions.

    Poništi
  2. 29. sij

    Many challenges still remain. Perhaps the biggest is how to satisfy causal constraints when generating counterfactuals, so that they are actionable. E.g., it is impossible to change one's education without aging. Some preliminary work 3/3

    Prikaži ovu nit
    Poništi
  3. 29. sij

    Rather than approximating the model, counterfactuals are promising since they always convey an accurate picture about the ML model. We outline our method for generating multiple diverse CFs in our paper: with & Ramaravind Mothilal

    Prikaži ovu nit
    Poništi
  4. 29. sij

    Excited to release a library for explaining ML models using counterfactual examples. The idea is simple--explain a prediction by the minimal changes that would lead to a different outcome--but turns out generating a diverse set is challenging. 1/3

    Prikaži ovu nit
    Poništi
  5. 5. sij

    Thank you for attending! For those interested, the full set of slides are at

    Poništi
  6. 18. pro 2019.

    Really like the idea of supplementary analyses . and I have implemented some of them in our library DoWhy, more to come!

    Poništi
  7. 14. pro 2019.

    Appropos Bengio's talk at , here's a great summary on applying causal reasoning for machine learning problems like generalization and adversarial robustness by

    Poništi
  8. 12. pro 2019.

    Excited to share that & I are writing a book on in computing systems. Summary: Machine learning needs causality, and causal inference needs ML methods. Here's the 1st chapter: We'll be sharing drafts online,would love your feedback

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

    Another academic sibling presenting his talk on cross-cultural differences in online mental health communities' use. Work /w

    Prikaži ovu nit
    Poništi
  10. 22. lis 2019.

    Looks like an interesting competition on causal discovery

    Poništi
  11. proslijedio/la je Tweet
    26. ruj 2019.
    Poništi
  12. proslijedio/la je Tweet
    9. kol 2019.
    Poništi
  13. proslijedio/la je Tweet

    Microsoft is expanding its fellowship program with the new Microsoft Investigator Fellowship. We're seeking PhD researchers of all disciplines who plan to make an impact through teaching and research using the Microsoft Azure platform. Submit now:

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

    We are very excited to announce the 3rd NeurIPS workshop on Machine Learning for the Developing World (ML4D)! This year’s theme is ‘Challenges and Risks’: What can go wrong with ML4D technologies? How do we tackle this?

    Prikaži ovu nit
    Poništi
  15. proslijedio/la je Tweet
    25. srp 2019.

    Our day2 summer school on HCAI ended with the presentations and discussions on HCML from Munmun De Choudhury (), Amit Sharma (), Kalika Bali (), and the MakerGhat team (@AzraIsmail1 & ).

    Prikaži ovu nit
    Poništi
  16. 25. srp 2019.

    Thank you for organizing this wonderful summer school bringing together HCI and ML. Enjoyed the conversations on human-centered ML. Slides from my talk:

    Poništi
  17. 24. srp 2019.

    Yep, and low-resource environments more broadly. Great to see conferences like COMPASS and ICLR happening in Africa this year.

    Poništi
  18. 21. srp 2019.

    Really enjoyed listening to on the podcast about machine learning and societal impact. Cool examples on how prediction does not always help with making decisions, and how interpretability and fairness are intertwined with social good.

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

    Attending the this year? I'm having fun organizing the "Machine Learning in Resource Constrained Environment" deep dive. We have some amazing speakers lined up, and also want to profile work you have done on ML at the edge. Submissions open for spotlight talks!

    Prikaži ovu nit
    Poništi
  20. proslijedio/la je Tweet
    15. srp 2019.

    Awesome thread about biotech and causality/correlation.

    Poništi
  21. proslijedio/la je Tweet
    11. srp 2019.

    Our library has an awesome new API, courtesy of ! Main idea: The do-sampler API transforms a dataframe of observational data into a dataframe that estimates the interventional distribution. Read all about it:

    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:

    ·