Akanksha Atrey

@akanksha_atrey

Full-time PhD student, part-time human — UMass Amherst — causal inference | explainable AI | blockchain | privacy.

Amherst, MA
Vrijeme pridruživanja: kolovoz 2014.

Tweetovi

Blokirali ste korisnika/cu @akanksha_atrey

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

  1. Prikvačeni tweet
    20. pro 2019.

    Very excited to announce that our paper titled “Exploratory Not Explanatory: Counterfactual Analysis of Saliency Maps for Deep RL” has been accepted at !

    Prikaži ovu nit
    Poništi
  2. proslijedio/la je Tweet
    1. velj
    Odgovor korisniku/ci

    Friendly counter-point: celebrate effort, not outcome ('success'?). Celebrate the submission of an article for the first time, not the editor's decision. Celebrate submitting the grant. Celebrate your student having finished data collection, not their p-value.

    Poništi
  3. proslijedio/la je Tweet
    14. sij

    Academic advice is riddled with survivorship bias

    Prikaži ovu nit
    Poništi
  4. proslijedio/la je Tweet
    2. sij

    "The AI community will place more value on performance beyond accuracy in 2020, like the amount of power it takes to create a model, how output can be explained to humans, and how AI can better reflect the kind of society people want to build."

    Poništi
  5. proslijedio/la je Tweet
    1. sij

    I want to give some perspective for people not from Australia. More Aussie land is currently burning than exists in the entire country of Belgium. The smoke is causing breathing problems in New Zealand, 2,000km away. Half a billion animals have been killed. Eight people are dead.

    Prikaži ovu nit
    Poništi
  6. proslijedio/la je Tweet
    1. sij

    I wouldn't recommend basing your career on currently popular trends, since these are likely to change by the time you graduate. Instead, figure out what questions/problems most fascinate you and how to make a career out of those. Define your own fields if you have to.

    Poništi
  7. 30. pro 2019.

    This promotes diversity and is one reason our lab (KDL) has consistently had a 1:1 men to women ratio.

    Prikaži ovu nit
    Poništi
  8. 30. pro 2019.

    My advisor, David Jensen, is one of the few I know who admits students to his lab based on their potential and not on what they have achieved. The key question is: if x student had the same opportunities as y student(s), where could they have been?

    Prikaži ovu nit
    Poništi
  9. 30. pro 2019.

    PhD is supposed to teach you how to become an independent researcher, not expect you to be one before you even start. Frustrated by this culture in ML/AI.

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

    Looking for a postdoc to work on topics in fairness+bandit+RL, as part of my research group at IEOR and DSI at Columbia. If interested, send me your CV with a list of three references, and any additional information. Feel free to DM. Please retweet. Be a matchmaker!

    Prikaži ovu nit
    Poništi
  11. 23. pro 2019.

    Seeing some on working over the holidays to meet upcoming deadlines...who in their right mind sets these early to mid Jan deadlines and why? It's hard to believe that those deadlines couldn't be moved up to late Jan/early Feb.

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

    You used to be able to get a faculty position with just a couple publications. Nowadays, having papers helps get into grad school. But using them as an edge to get into college? What's next, kindergarten placement based on h-index?

    Poništi
  13. 20. pro 2019.

    Realizing that being in a PhD program can often be a privilege. Folks that have loans, family responsibilities or other financial responsibilities can’t afford to do a PhD even if they want to.

    Poništi
  14. proslijedio/la je Tweet
    20. pro 2019.

    Super excited for this first published work that uses - and we have a big release coming in the new year, so follow now to get notifications in your github feed!

    Poništi
  15. 20. pro 2019.

    This is joint work with @kclary92 and David Jensen. Wouldn’t have been possible without their hard work. Thanks to , , and many more for the support and useful feedback throughout the process.

    Prikaži ovu nit
    Poništi
  16. 20. pro 2019.

    The paper presents (1) a survey of how saliency maps (SMs) are used in the deep RL community, (2) a method to evaluate the inferences made from SMs about agent behavior, and (3) an experimental evaluation to measure how well SMs correspond to semantic-level inferences of humans.

    Prikaži ovu nit
    Poništi
  17. proslijedio/la je Tweet
    17. pro 2019.

    One theme I noticed at was the need for the ML community to move beyond our narrow focus on prediction. I heard this in 3 very different talks. [Thread]

    Prikaži ovu nit
    Poništi
  18. 18. pro 2019.
    Poništi
  19. proslijedio/la je Tweet

    With both House votes official, President Trump has been IMPEACHED on both counts against him: Abuse of Power and Obstruction of Congress.

    Poništi
  20. 16. pro 2019.

    “This is a SOTA deep learning paper. Don’t try this at home.” — a deep friend. 😆

    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:

    ·