Victor Veitch

@victorveitch

machine learner, statistician, effective altruist, radical preachy vegan

Vrijeme pridruživanja: svibanj 2013.

Tweetovi

Blokirali ste korisnika/cu @victorveitch

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

  1. Prikvačeni tweet
    6. lip 2019.

    We want to estimate a causal effect by adjusting for (gnarly) covariates using a neural net. How should we design the neural net? Answer: translate classical statistical results to neural net design! Works great. with and

    Prikaži ovu nit
    Poništi
  2. proslijedio/la je Tweet
    prije 5 sati

    Denied nuclear, Japan's utility companies rush to build massive new coal plants.

    Poništi
  3. proslijedio/la je Tweet
    1. velj

    I'm not sure if the field is in fact self correcting, if its true then thats great news. But I agree with the rest and the field needs to go in the direction of self correcting. Constantly chasing deadlines without the ability to think is not good for science health or happiness.

    Poništi
  4. proslijedio/la je Tweet
    23. sij

    Pre-2010, 50g of crack-cocaine got you a mandatory minimum 10 years in prison. Post-2010, that threshold was raised to 280g, which resulted in this .... uhhh suspicious drug seizure graph. h/t

    pre 2010, pretty reasonable power law graph
    post 2010, pretty reasonable power law graph, except for a huge spike right at the 280g threshold
    Prikaži ovu nit
    Poništi
  5. proslijedio/la je Tweet
    25. sij

    a true o henry story of transition....i started T so i could sing the Javert parts in confrontation....and yet now i can no longer sing the higher valjean parts in confrontation...damn these stars

    Poništi
  6. proslijedio/la je Tweet
    21. sij

    I have post-doc openings in my lab. Do you want to use empirical methods, design experiments, and collaborate with both large and small tech firms for research and social impact? Have interest in ed tech, fin tech, charitable giving, and social sector effectiveness? Apply! 1/3

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

    Reminders of deadlines: Papers: abstracts due 30 Jan, papers due 6 Feb Workshops: due 14 Feb Tutorials: due 21 Feb details on each: Please submit (and circulate the calls!)

    Prikaži ovu nit
    Poništi
  8. proslijedio/la je Tweet
    23. sij
    Odgovor korisniku/ci

    Oh you mean a no hidden layer deep neural network with only 1 sigmoidal output neuron? Why didn’t you just say so - instead of using obscure statistical jargon like “logistic regression?” :)

    Poništi
  9. 23. sij

    This would be a big improvement to the practice of science in the US

    Poništi
  10. proslijedio/la je Tweet
    17. sij

    Hey everyone, pls come hang out and talk about social science!!

    Poništi
  11. proslijedio/la je Tweet
    10. sij

    Willie Nash was at the county jail on a misdemeanor when he asked an officer to charge the cell phone he walked in with. Instead, he got a 12-year prison term for bringing the phone inside. And the judge told him: "consider yourself fortunate." 1/

    Prikaži ovu nit
    Poništi
  12. 12. sij

    I'd like something I can monitor in a similar fashion to accuracy. Accuracy, ofc, doesn't work because the ground truth classifier would just predict 0 for every feature.

    Prikaži ovu nit
    Poništi
  13. 12. sij

    Is there a standard(ish) metric for calibration useful in the following setting? Unknown to you: Objects with features in set 'A' are 1 wp 30% Objects with features in set 'B' are 1 wp 10% Question: do I have a good model of P(Y=1 | features) ?

    Prikaži ovu nit
    Poništi
  14. 8. sij

    Every once in a while I find a solution to a pretty common problem that would take ~half a page to write down. What's the best place to put stuff like that? (e.g., a blog post would be appropriate, but might never be indexed to be found by other stuck people)

    Poništi
  15. 7. sij

    Caveat (the bar for nerds) is my favorite venue in NYC---highly recommend checking out their stuff :)

    Poništi
  16. proslijedio/la je Tweet
    7. sij

    Meta-reviewer boldly claims that our entire theorem is wrong by providing a "counterxample" wherein they claim S=[1 0; 0 -2] to be a positive definite matrix, while denouncing our claims to be false!! 🤯Are you sure you want to be meta-reviewing? Please be more considerate.

    Poništi
  17. proslijedio/la je Tweet
    1. sij

    I will go on record to say that I think it's crazy people want to stop the SAT from being factored into US college admissions, when the actually-existing alternative admissions criteria are things like this.

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

    My research group (located at Cornell Tech campus in NYC) is looking to recruit a postdoc to work on topics related to causal inference, fairness in ML, and sequential decision making (bandits+RL). Positions are renewable (1-2 years). Please retweet to spread the word. 🙏

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

    Here at we are looking for philosophy faculty specializing in topics related to ethics & computing. One of the first two searches for the new College of Computing.

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

    Visit our poster (#232) tomorrow (Thursday) during the evening poster session from 5 to 7 pm to learn about our work bounding the expected generalization error of SGLD using information theoretic tools.

    Poništi
  21. 11. pro 2019.

    If you're at , come talk to us at the Human-Aligned AI Social on Thursday :)

    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:

    ·