Twitter | Pretraživanje | |
Frans Zdyb
22
Tweetovi
148
Pratim
31
Osobe koje vas prate
Tweetovi
Frans Zdyb 23. sij
Srsly, if they'd make a movie about this guy it would make James Bond look realistic in comparison.
Reply Retweet Označi sa "sviđa mi se"
Frans Zdyb 14. sij
Odgovor korisniku/ci @ToKTeacher @DavidDeutschOxf i 2 ostali
The unique feature that cpu's and brains have in common is that they accept arbitrary data and programs as input and reliably produce the correct output. But brains are even more unique because they program themselves (to some extent).
Reply Retweet Označi sa "sviđa mi se"
Frans Zdyb 14. sij
Odgovor korisniku/ci @ToKTeacher @DavidDeutschOxf i 2 ostali
I like the post, but I think it's incomplete. Your argument says brains are physical systems, and all physical systems can be simulated by a Turing machine. But some systems are better understood as computers than others (a cpu vs a piece of quartz). (1/2)
Reply Retweet Označi sa "sviđa mi se"
Frans Zdyb proslijedio/la je tweet
David Chapman 9. sij
. is an extraordinary, massive, persistent force of nature, like Jupiter’s Great Red Spot or something. Read his decade in review and be gobsmacked:
Reply Retweet Označi sa "sviđa mi se"
Frans Zdyb 28. pro
Odgovor korisniku/ci @michael_nielsen
No metal in the responses? Tesseract - Altered State Animals as Leaders - Joy of Motion
Reply Retweet Označi sa "sviđa mi se"
Frans Zdyb 5. stu
Odgovor korisniku/ci @acapellascience
Information as a concept only makes sense when one physical system encodes it s.t. another can decode it. If you consider the entire universe, it's impossible to specify a decoder, and the question dissolves.
Reply Retweet Označi sa "sviđa mi se"
Frans Zdyb 25. lis
Odgovor korisniku/ci @amine_kechaou @GaryMarcus i 3 ostali
Douglas Hofstadter wrote an essay about Google Translate that explains it
Reply Retweet Označi sa "sviđa mi se"
Frans Zdyb proslijedio/la je tweet
Rohin Shah 21. lis
Excited to share our work: collaboration requires understanding! In Overcooked, self-play doesn't gel with humans: it expects them to play like itself. (1/4) Demo: Blog: Paper: Code:
Reply Retweet Označi sa "sviđa mi se"
Frans Zdyb 14. lis
Odgovor korisniku/ci @FateOfTwist_
Alright, I'm pretty sure I'm not getting your point, maybe let's just leave it at that :)
Reply Retweet Označi sa "sviđa mi se"
Frans Zdyb 14. lis
Scientific realism is bad philosophy?
Reply Retweet Označi sa "sviđa mi se"
Frans Zdyb 12. lis
Odgovor korisniku/ci @acapellascience @smartereveryday
Why do we still see eye floaters though? Seems like an easy thing to remove in post processing.
Reply Retweet Označi sa "sviđa mi se"
Frans Zdyb proslijedio/la je tweet
Future of Life 9. lis
Odgovor korisniku/ci @elonmusk
We've interviewed Stuart about his new book here:
Reply Retweet Označi sa "sviđa mi se"
Frans Zdyb 4. lis
Odgovor korisniku/ci @ChadScherrer
On second thought, maybe we assume that the posterior is low-entropy enough that MC-expectations over it are that much more robust than with the prior?
Reply Retweet Označi sa "sviđa mi se"
Frans Zdyb 4. lis
Odgovor korisniku/ci @ChadScherrer
I'm thinking one could take an approach for ML and adapt it (e.g. ). I love the idea of Soss.jl, btw. It looks like we're interested in many of the same things! I've been experimenting with ASTs and programmatic transformations of Pyro models for example.
Reply Retweet Označi sa "sviđa mi se"
Frans Zdyb 3. lis
Odgovor korisniku/ci @ChadScherrer
I mean, ML is integrating the product of likelihood and prior over all parameters; PP is integrating the product of likelihood and posterior over the same space. There are countless papers on approximating ML in various ways, but the standard prescription for PP is just sampling.
Reply Retweet Označi sa "sviđa mi se"
Frans Zdyb 3. lis
Why do people estimate the predictive posterior by regular sampling when it's exactly as hard as computing marginal likelihood? Please tell me I'm missing something obvious.
Reply Retweet Označi sa "sviđa mi se"
Frans Zdyb 27. ruj
Great to meet ML people in the industry :) so long and thanks for all the swag!
Reply Retweet Označi sa "sviđa mi se"
Frans Zdyb proslijedio/la je tweet
Steven Strogatz 25. ruj
"MIT Mathlets" = These are terrifically visual, interactive applets for learning about differential equations and related parts of math. Very intuitive and easy to use, and beautifully designed by Hubert Hohn, an artist who took my chaos course at MIT
Reply Retweet Označi sa "sviđa mi se"
Frans Zdyb 15. kol
Odgovor korisniku/ci @DRMacIver
Surfing Uncertainty The Moral Animal Mastering the Core Teachings of the Buddha (1) The contents of consciousness are mental fabrications, (2) whose function is not to reflect reality, but be adaptive. (3) If you focus really hard on those contents, you'll see it for yourself.
Reply Retweet Označi sa "sviđa mi se"
Frans Zdyb proslijedio/la je tweet
Max Welling 20. tra
I wrote this response to Rich Sutton's "The Bitter Lesson"
Reply Retweet Označi sa "sviđa mi se"