Robert Dionne

@robertsdionne

Software Engineer interested in Machine Learning and Natural Language Processing and Generation; MFA graduate from NYU Game Center.

New York, NY
github.com/robertsdionne/…
Joined June 2009
170 Photos and videos Photos and videos

Tweets

You blocked @robertsdionne

Are you sure you want to view these Tweets? Viewing Tweets won't unblock @robertsdionne

  1. Pinned Tweet
    30 Jun 2016

    My summary of the Deep Residual Networks ICML 2016 tutorial:

    3 retweets 23 likes
  2. Retweeted
    Jun 9
    Replying to

    Ran a 10 layer NN and viewed the activation distributions. Here are the results for SELU and ReLU respectively.

    4 replies 27 retweets 71 likes
  3. Retweeted
    Jun 9
    Replying to

    This is the magic SELU activation function.

    9 replies 97 retweets 280 likes
  4. Retweeted
    Jun 2

    Brilliant new typeface by Liron Lavi Turkenich uses perceptual principles to hybridize Arabic & Hebrew scripts.

    3 replies 68 retweets 132 likes
  5. Retweeted
    May 17

    11.5 petaflops... Now google "number of synaptic connections in human brain": Impressive !

    2 retweets 5 likes
  6. Retweeted
    May 3

    Hey, if you're a journalist getting hit with random Google Doc invites right now, do not click. Forward phishing attacks to cooperq@eff.org

    120 replies 2,578 retweets 1,698 likes
  7. Retweeted
    Apr 27

    A New Type of Neurons for Machine Learning

    5 retweets 8 likes
  8. Retweeted
    Apr 25
    Replying to

    Lefsetz piece now online:

    3 replies 33 retweets 67 likes
  9. Retweeted
    Apr 18

    "Representing Sentences as Low-Rank Subspaces," Mu et al.:

    4 retweets 7 likes
  10. Retweeted
    Apr 21

    Here we go.

    45 replies 201 retweets 633 likes
  11. Retweeted
    Apr 18

    New blog post on acapella extraction with ConvNets in –with samples + code! Happy to finally share this 😁

    5 replies 9 retweets 16 likes
  12. Retweeted
    Apr 19

    "Learning to Fly by Crashing," Gandhi et al.:

    12 retweets 19 likes
  13. Retweeted
    Apr 19

    We've open sourced the Differentiable Neural Computer! Built with and . Available here:

    10 replies 514 retweets 919 likes
  14. Retweeted
    Apr 10

    Mathematicians have discovered how the universal patterns behind innovation arise - via

    10 retweets 39 likes
  15. Retweeted
    Apr 9

    overbook and decided to force random passengers off the plane. Here's how they did it:

    7,894 replies 168,292 retweets 150,451 likes
  16. Retweeted
    Mar 30

    Evolution Strategies: embarrassingly simple, distributed, gradient-free optimisation review+ thoughts on extensions:

    5 replies 129 retweets 310 likes
  17. Retweeted
    Mar 28

    GCI solved nearly 3 yrs ago. Proof uses nothing more than undergraduate-level stats and calculus!

    10 retweets 23 likes
  18. Retweeted
    Mar 24

    New blog post from OpenAI on "Evolution Strategies as a Scalable Alternative to Reinforcement Learning" w00t!!

    9 replies 206 retweets 406 likes
  19. Retweeted
    Mar 15

    Really nice paper from connecting data-noising for neural-LM with traditional smoothing techniques.

    18 retweets 43 likes
  20. Retweeted
    Mar 15

    Nuclear physicists have spent years analyzing these videos in the hopes of wringing every last bit of data they can

    Lawrence Livermore Nat'l Lab put 60 videos of nuclear explosions on YouTube. Not relaxing. (via )
    41 retweets 74 likes
  21. Retweeted
    Mar 14

    Wow! New search theory is competitive with A* or outperforms it significantly. Near-Optimal Bidirectional Search

    6 replies 225 retweets 460 likes

@robertsdionne hasn't Tweeted yet.

Loading seems to be taking a while.

Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.

    You may also like

    ·