robrenaud

42 ms·

481.
by robrenaud 14y ago
I have one Chinese friend who married a Chinese American and had a child with her. He then got a job opportunity in Shanghai and took it, leaving her in the US with the child because the food safety is so bad in China. Knowing nothing abo
482.
by robrenaud 14y ago
You could also ask many humans and average their response to obtain a golden label, and see how well any particular human agrees with the average. If there is a lot of variance in the human answers, then it's possible for a machine to have
483.
by robrenaud 14y ago
I know these test names might seem insane. And I personally hate the Java FactoryFactory culture of over-engineering. But, the super long test class names are actually pretty nice. I've personally never gone quite that crazy with the nam
484.
An academic statistician's view on Nate Silver and betting one's predictions(normaldeviate.wordpress.com)
2 points by robrenaud 14y ago|0 comments
485.
by robrenaud 14y ago
spoj.pl is very similar to acm.uva.es, but also has a much wider array of supported languages.
486.
by robrenaud 14y ago
I interview for Google. I gave my question to an algorithms professor and he didn't get the optimal answer. Needless to say, I don't expect candidates to get the optimal answer.
487.
by robrenaud 14y ago
Coursera is awesome, don't get me wrong. I am on my fourth Coursera class (if you count the pre-Coursera ML class, and Quantum Mech that I gave up on a couple weeks in), and I haven't had any real interest in anything Udacity has had to of
488.
by robrenaud 14y ago
He also recalls being 5 years older than his birth certificate says.
489.
by robrenaud 14y ago
Work as a search quality engineer at Google and you do pretty much all of that. Except for running the clusters[1], I've done pretty much all of those steps myself. I started with a nice statistical idea, built some simple models, played
490.
by robrenaud 14y ago
> That cheating is so easy on Coursera, "give me an A" is precisely what's going on for people that are there, not for the learning experience, but for an A. Maybe the 'A' or nice cert is there for some people, but no one that I know pe
491.
by robrenaud 14y ago
Isn't NFL utter crap? When you average an learning algorithms performance over a whole bunch of domains that _NATURE WILL NEVER GENERATE_, all algorithms are equally bad. Paying attention to the theorem is mostly defeatist and counter-produ
492.
by robrenaud 14y ago
I am not going to buy a lottery ticket because the super unlikely best case is really good. For the same reason, I am not going to avoid a hash table because the super unlikely worst case is kind of bad. You never have 1000 items in each b
493.
by robrenaud 14y ago
> As a developer there's a lesson that needs reinforcing every-so-often. What better way to reinforce it than to see a tech giant make some of the mistakes lesser companies make: If you can at all help it, don't base your product on som
494.
by robrenaud 14y ago
To me, broker fees in NYC seem so crazy. Can someone justify them? Is opening up a few doors really worth 3k?
495.
by robrenaud 14y ago
You missed the "edit screenshot in photo editor to highlight and black out info" part. Having used this internally, I can say yes, this tool flows very well. I don't know if we do this, but you could imagine clustering feedback by the highl
496.
by robrenaud 14y ago
> Rodrigo is an MIT graduate who writes compilers in his spare time. He is a core contributor to Haskell and wrote a few very well known Python packages. He can generally write very solid code that's readable and handles edge cases beau
497.
by robrenaud 14y ago
Python's hashes for small integers never collide. Python 2.6.5 (r265:79063, Apr 16 2010, 13:57:41) >>> map(hash, range(10)) [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]
498.
by robrenaud 14y ago
You are assuming that having takedown notices does not correlate with a lack of quality. I don't have any inside information about this change in particular, but I'd guess that is a bad assumption. I'd bet that indeed, this signal is usef
499.
How to choose colors procedurally(devmag.org.za)
7 points by robrenaud 14y ago|0 comments
500.
by robrenaud 14y ago
Does anyone know how the nucleus destruction happens? Is there any analogy to regularization/sparsification in machine learning? Is there some kind of process that destroys the nucleus of the least useful neurons?
501.
by robrenaud 14y ago
I was impressed that random forests did so well for the irrelevant feature detection, given that they know nothing about gaussians. Though IIRC, you've used them to win Kaggle competitions, so maybe you already know their power.
502.
by robrenaud 14y ago
This isn't making a permanent record of casual encounters. It's putting apartment listings on a map. People listing apartments tend to want it to be widely known.
503.
by robrenaud 14y ago
Andrew Ng has set many state of the art results on various data sets using similar approaches as the one described in the paper. Here is a reasonably approachable talk he gave about it. http://www.youtube.com/watch?v=ZmNOAtZIgIk
504.
by robrenaud 14y ago
An image is a lot more complicated than a pair of ids and a rating. Counting the number of rows in the training database is misleading. I can build a reasonable dataset for a prediction task from a set of 100M rows from a database that I
505.
by robrenaud 14y ago
There is a lot of use in say, complaining that a way you are going about a problem will work, but seems too damn hard. Say for example, I see a 4 hour solution to a problem that should really be solved in 30 minutes. Often coworkers who
506.
by robrenaud 14y ago
Mostly I agree with you. But Google does have a strong C++/Java/Python bias. Even if it hasn't touched you personally, it is very likely that a new Googler will be using one of those 3 languages as her main language.
507.
by robrenaud 14y ago
Actually, I think this might piss off the government more. Imagine Google could perfectly predict what would be censored before even querying. Now you can search for related, uncensored terms and find the information you want without havin
508.
by robrenaud 14y ago
The commands module is now deprecated by the overly complicated subprocess module, but content = commands.getoutput(x) is much better than os.system(x + ' > foo') content = open('foo', 'r').read()
509.
by robrenaud 14y ago
I've heard an even crazier idea. Have colleges charge a percentage of a student's future income. This aligns economic incentives for the student and the college.
510.
by robrenaud 14y ago
They came from NY.

More ›