Anish Athalye

@anishathalye

grad student @ mit

Cambridge, MA
anishathalye.com
Vrijeme pridruživanja: travanj 2012.
16 fotografija ili videozapisa Fotografije i videozapisi

Tweetovi

Blokirali ste korisnika/cu @anishathalye

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

  1. proslijedio/la je Tweet
    prije 12 sati

    In January, , , and I ran a short class at on topics we think are missing in most CS programs — tools we use every day that everyone should know, like bash, git, vim, and tmux. And now the lecture notes and videos are online!

    8 replies 131 proslijeđeni tweet 462 korisnika označavaju da im se sviđa
    Poništi
  2. 30. pro 2019.

    distributed systems aesthetics (from a WIP visualizer for partial linearizations of concurrent histories)

    3 korisnika označavaju da im se sviđa
    Poništi
  3. 20. pro 2019.

    Made a minimalist LaTeX beamer presentation theme: (to pair with the Gemini poster theme: )

    1 reply 4 proslijeđena tweeta 19 korisnika označava da im se sviđa
    Poništi
  4. 12. pro 2019.

    Here's one of my favorite results -- it's what happens when you solve a figure using gradient descent. The TensorFlow-based solver is probably useless, but its results are oddly satisfying.

    3 proslijeđena tweeta 43 korisnika označavaju da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  5. 12. pro 2019.

    I've been working on this new system for graphic design, where drawings are specified in terms of constraints and solved using an SMT solver. Some interesting and amusing results have come out of these experiments:

    20 replies 127 proslijeđenih tweetova 547 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  6. proslijedio/la je Tweet
    4. stu 2019.

    Confident Learning for finding label errors in datasets and learning with noisy labels is finally here! Read about it: Reddit: Paper: NEW Python Package:

    25 proslijeđenih tweetova 62 korisnika označavaju da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  7. proslijedio/la je Tweet
    11. kol 2019.

    insta-meme

    565 proslijeđenih tweetova 2.045 korisnika označava da im se sviđa
    Poništi
  8. 10. kol 2019.

    Talking on improving security using the age-old wisdom of "turning it off and on again," applied to hardware wallets. New hardware designs, formal verification, towards principled security... (w/ demos + code) Sun 1pm, Paris Theatre (Track 4)

    1 reply 6 korisnika označava da im se sviđa
    Poništi
  9. proslijedio/la je Tweet
    18. srp 2019.

    We are looking for motivated volunteers to form the artifact evaluation committee at . If you are a grad student or researcher and think reproducible science is important, this is your chance to help towards that goal! Please sign up:

    Dear systems people - I'm excited to announce that for the first time SOSP will be organizing artifact evaluation (led by and ). If you are an early career researcher, postdoc, or senior grad student, please apply to join us!
    1 reply 25 proslijeđenih tweetova 36 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  10. 29. tra 2019.

    ML Tidbits, started by two PhD students at MIT, is bringing understandable explanations of AI and ML to the public. More at

    Check out our first two videos on the lifecycle and different types of ML!
    1 proslijeđeni tweet 4 korisnika označavaju da im se sviđa
    Poništi
  11. proslijedio/la je Tweet
    10. tra 2019.

    3 years ago MIT grad student Katie Bouman led the creation of a new algorithm to produce the first-ever image of a black hole. Today, that image was released. More info: 2016 story:

    1.271 reply 58.018 proslijeđenih tweetova 141.731 korisnik označava da mu se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  12. proslijedio/la je Tweet
    21. velj 2019.

    Announcing reproducible benchmarks for all Keras- and pre-trained models. rules ResNets. Keras rules InceptionNets. Blog post: GitHub:

    1 reply 30 proslijeđenih tweetova 94 korisnika označavaju da im se sviđa
    Poništi
  13. proslijedio/la je Tweet
    18. velj 2019.

    We published guidelines for evaluating adversarial robustness (). This is a living document: please feel invited to join & share your thoughts. By Nicholas Carlini

    31 proslijeđeni tweet 89 korisnika označava da im se sviđa
    Poništi
  14. proslijedio/la je Tweet
    9. velj 2019.

    I just re-recorded the Shell and Scripting lecture from 👇 with a proper screencast and microphone setup. Should be much easier to follow now!

    All of the videos from the MIT lecture series , , and I ran over the course of January on how to make the most of the tools that productive developers use that are now finally available \o/
    3 proslijeđena tweeta 27 korisnika označava da im se sviđa
    Poništi
  15. proslijedio/la je Tweet
    2. velj 2019.

    All of the videos from the MIT lecture series , , and I ran over the course of January on how to make the most of the tools that productive developers use that are now finally available \o/

    1 reply 25 proslijeđenih tweetova 85 korisnika označava da im se sviđa
    Poništi
  16. proslijedio/la je Tweet
    29. sij 2019.

    For the past few weeks, , , and I have been running a brief lecture series at for IAP that teaches you how to make the most of the tools that productive developers use, and the first few lecture videos are now up at !

    1 reply 15 proslijeđenih tweetova 50 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  17. 29. pro 2018.

    Simpler human-readable labels for ImageNet: (thought I'd share in case it's useful to anyone)

    7 korisnika označava da im se sviđa
    Poništi
  18. proslijedio/la je Tweet
    15. lis 2018.

    Excited to announce Piano Genie, an intelligent controller that allows anyone to improvise on the piano! This was my internship project at with and . Blog post with more information:

    26 replies 742 proslijeđena tweeta 2.211 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  19. proslijedio/la je Tweet
    30. kol 2018.

    Sketchnote summary of "Obfuscated gradients give a false sense of security", and Nicholas Carlini, ICML 2018

    5 proslijeđenih tweetova 15 korisnika označava da im se sviđa
    Poništi
  20. proslijedio/la je Tweet
    27. srp 2018.

    We analyze the Adversarial Logit Pairing defense (training objective, loss landscape, and adv robustness). Our results show that released ALP models are not robust. Loss viz below! Paper: Code: (with )

    1 reply 11 proslijeđenih tweetova 21 korisnik označava da mu se sviđa
    Poništi

@anishathalye još nije objavio nijedan Tweet.

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:

    ·