gabrielgoh

11 ms·

61.
by gabrielgoh 10y ago
these fallacies are vague enough that most arguments can be twisted to be in violation of one or more of these and dismissed
62.
by gabrielgoh 10y ago
My experience points the other way. Surveys and textbooks are good in principle, but I rarely find myself using them in research. When you're looking to understand a very specific thing (e.g. what is ADMM? How fast does the conjugate g
63.
by gabrielgoh 10y ago
so did Kevin Kelly, in "The Inevitable". He stretches the electrical metaphor for about a chapter. "The AI on the horizon looks more like Amazon Web Services — ... This common utility will serve you as much IQ as you want but
64.
by gabrielgoh 10y ago
Find out in a week!
65.
by gabrielgoh 10y ago
i disagree with the first point. I'm working on a distill article with Chris and Shan, and the major draw for this has been impact. It seems very plausible that an article on distill has the potential to reach a far broader (and differ
66.
by gabrielgoh 10y ago
shoot me an email with your background etc. I'm not aiming for this batch actively but i wouldn't mind a chat. Google my name to get my website an email. it should be the first result
67.
by gabrielgoh 10y ago
im not quite sure what you're suggesting. there needs to be a source of ground truth to inform the robot if what is being done is right. And actually having a physical robot arm for this would be way too slow
68.
by gabrielgoh 10y ago
it is for simulation, as you say - but this is a critical for employing reinforcement learning. The internal (inverse) model will of course, be based on a simplified system, like a DNN
69.
by gabrielgoh 10y ago
I've wanted to break into robotics from machine learning for a long time, but I haven't found a good entry point for the problem. It seems like a rather large vertical cliff I have no way of scaling. One of the handholds I'd
70.
by gabrielgoh 10y ago
reinforcement learning, the most promising approach for robotics today, doesn't require a lot of data to train. Or rather the data required to train it can be synthesized from the problem domain itself. Access to compute and hardware,
71.
by gabrielgoh 10y ago
the article is painfully verbose, and doesn't even get into what the objective function is! (it just gives a high level description of what dynamic programming is) And the paper is worse, droning on the idea for 82 pages!! The idea is
72.
by gabrielgoh 10y ago
probably our entire conception of psychology and psychiatry will go a radical change. To think we still do electroshock therapy (lets induce a seizure in the brain and hope for the best!), have no idea how it works, and this is still consid
73.
by gabrielgoh 10y ago
What you're forgetting is radioactive underpants were considered kooky even in their time - these were the homeopaths and creationists of their time. I don't think our common-sense beliefs as humans haven't changed as radical
74.
by gabrielgoh 10y ago
principle of parsimony
75.
by gabrielgoh 10y ago
there is indeed. my comment was truly superfulous
76.
by gabrielgoh 10y ago
does this have cuda support?
77.
by gabrielgoh 10y ago
Interesting! I've never heard climate science explained this way before. Could you point me to a few papers which expound on this view? I'd be interested to see the mathematical models behind this.
78.
by gabrielgoh 10y ago
A few misconceptions I should correct in this comment. - There is plenty of existing technology for handling non-differentiable function. Functions like the absolute value, 2-norm, and so on have a generalization of the gradient (the subgra
79.
by gabrielgoh 10y ago
i see ray kurtzwell managed to sneak his name in as the last author. heh
80.
by gabrielgoh 10y ago
naw, it's a 3-book long novel and none of the story would make sense without the first 2. The first 2 are good too
81.
by gabrielgoh 10y ago
yeah, the book was pretty mind blowing. That book blew my mind the way reading Philip K Dick did when I was 16. I didn't think books could do that anymore. I recommend this book to anyone who wants a further exploration of the ideas me
82.
by gabrielgoh 10y ago
certainly up there with great titles like "can one hear the shape of a drum".
83.
Detecting Genre from Book Covers Using Deep Learning(arxiv.org)
47 points by gabrielgoh 10y ago|8 comments
84.
by gabrielgoh 10y ago
I've used both. Lua is powerful because of torch and it's GPU computing capabilities, but in terms of ease of use and modern syntax, Julia is lightyears ahead.
85.
A simple hypernetwork tutorial(github.com)
2 points by gabrielgoh 10y ago|0 comments
86.
A curated list of ICLR 2017 Submissions(smerity.com)
3 points by gabrielgoh 10y ago|0 comments
87.
by gabrielgoh 10y ago
Im not completely sold on his book being a textbook. It is incredible bit of research, I agree, but it still has rough edges and should be treated as such, a work in progress. I do not think this book should be the bible of causality, and
88.
by gabrielgoh 10y ago
This might be the fault of IBM's marketing, but it's completely unclear to me what the Watson API actually is. From the documentation, it seems to be a suite of algorithms which do standard NLP tasks, such sentiment analysis, topi
89.
by gabrielgoh 10y ago
the artifacts are explained in detail here http://distill.pub/2016/deconv-checkerboard/ more compute will enable higher resolutions and less artifacts, but it means training a bigger GAN
90.
by gabrielgoh 10y ago
By popular demand, I've added more pictures! https://open_nsfw.gitlab.io/more.html

More ›