tudorachim

8 ms·

31.
by tudorachim 16y ago
The most promising approaches for neural nets these days involve learning generative models using Restricted Boltzmann Machines (RBMs). Geoffrey Hinton at UToronto has some good seminal papers on these, like http://www.cs.toronto.edu/~hint
32.
by tudorachim 16y ago
Along the same tangent, Eigen has pretty much the nicest API for matrix math, in C++. Additionally, it is totally template based, and allegedly compiles to some highly optimized code. Even without that, it's a huge step up from the likes of
33.
by tudorachim 16y ago
Hey, I think my original post was totally misinterpreted. I was referring to the instance near the beginning where Arrington drove the robot forward, directly into Dan Casner, a few times. That mistake will almost certainly be repeated by e
34.
by tudorachim 16y ago
That's some pretty pathetic collision avoidance, given that the website advertises "Its built-in guidance system takes care of the rest by avoiding furniture & people..."
35.
by tudorachim 16y ago
There is a much more straightforward and principled way: the Metropolis algorithm ( http://en.wikipedia.org/wiki/Metropolis%E2%80%93Hastings_alg... ). You are basically sampling from the distribution of n-gram likelihoods by following a mar
36.
by tudorachim 16y ago
I've never heard of a school whose abbreviation is csun; thus, this list probably has a bunch of errors. Not worth the read.
37.
by tudorachim 16y ago
I think the main advantage is that this RNG gives you the same results as a crypto-strong PRNG, but much faster. That said, I didn't see any benchmarks on their site.
38.
by tudorachim 16y ago
The claim originally came from the Cambridge Handbook of Expertise and Expert Performance ( http://www.amazon.com/Cambridge-Expertise-Performance-Handbo... ).
39.
by tudorachim 16y ago
Yes, but it only works for managing X apps, not the standard ones OSX ships with.
40.
by tudorachim 17y ago
Although you may be able to bypass their protection, even if it's a simple cookie authentication, the whole operation will probably not be a failure. Presumably most people they are targeting, who might be willing to pay for a subscription,
41.
by tudorachim 17y ago
If I have an account and am taking a test that I've created, you should provide a link back to the dashboard after I've submitted so I can check how I did. Currently, I'm taken to the the "enter a password" screen and have to manually go ba
42.
by tudorachim 17y ago
Have you tried topcoder? It's the same as codility, but with cooler/harder problems.
43.
by tudorachim 17y ago
Wow, you actually need to create an account to order? I suppose there's no way I'm getting through that gauntlet before the $100k runs out. edit: In case you want to create an account in parallel while you're waiting, see https://www.spa
44.
by tudorachim 17y ago
After reading that post I'm still confused about why they want to "split off."
45.
by tudorachim 17y ago
I'm surprised http://en.wikipedia.org/wiki/Zobrist_hashing wasn't mentioned at all in the article.
46.
by tudorachim 17y ago
Presumably he instructed mathematica to approximate the roots to an error proportional to the resolution of the image; in that (likely) case the error doesn't have anything to do with the fractals.
47.
by tudorachim 17y ago
This is pretty sweet; it seems like a nicer way to do one-off pastebins than current sites.
48.
by tudorachim 17y ago
Aww, after I saw the integration examples I got my hopes up for a Show Steps for this... http://www.wolframalpha.com/input/?i=integrate%5Bexp%28-t%28...
49.
by tudorachim 17y ago
Wow! I found this part the most hilarious: "To calculate information about the influence of one weather station on another ... the code draws little coloured circles (actually little 32-sided polygons) on a virtual white screen and then goe
50.
by tudorachim 17y ago
http://en.wikipedia.org/wiki/Turing_machines , http://en.wikipedia.org/wiki/Quicksort , http://en.wikipedia.org/wiki/Dijkstra%27s_algorithm , http://en.wikipedia.org/wiki/Dynamic_programming , graph coloring algorithms, etc.
51.
by tudorachim 17y ago
"And, sure, the guys behind it had made some of the biggest contributions to computer science to date, including: UNIX, regular expressions, Plan 9 and even UTF-8!" It's a little far-fetched to say that two operating systems and a string co
52.
by tudorachim 17y ago
tar -xvf infers the filetype from the file; obviating the need for the smart untarring script. edit: Also, another useful thing is <command> | xargs -n1 -I{} <stuff>, which runs stuff once on every element of the output of c
53.
by tudorachim 17y ago
That does make sense, but the article is saying that people screwed up because they overthought their current strategies and played worse than normally, not because they overcorrected in the other direction and took more risks.
54.
by tudorachim 17y ago
0? I typed out a reply, but realized that it just reduces to the simple fact that 4chan is the collected ravings of thousands of people who use the anonymity they are given to act like idiots. Hopefully, it is not how they behave themselves
55.
by tudorachim 17y ago
Presumably because the author of the article found it good enough to write about.
56.
by tudorachim 17y ago
There are others, and more information here: http://scienceblogs.com/pontiff/2009/11/igon_value_problems....
57.
by tudorachim 17y ago
Hurr, engineers don't have feelings, so they don't understand emotional blah blah blah. I don't know how the author feels remotely qualified to write a book about a company with a famed engineering culture, when it's so obvious he hasn't co
58.
by tudorachim 17y ago
I'm not sure if this is your question, but divided differences are a way to do polynomial interpolation, with efficient updating to account for new data points: http://en.wikipedia.org/wiki/Newton_polynomial
59.
by tudorachim 17y ago
It's very slick. It would be nice to have a task automatically save its changes once you click somewhere else on the screen, instead of having to click on the "Save" button.
60.
by tudorachim 17y ago
Interesting. Which ones?

More ›