diab0lic

30 ms·

301.
by diab0lic 11y ago
I don't work on the recommendations side of things, but this is actually the ONE technical topic at Netflix that I don't believe I am allowed to discuss. Sorry!
302.
by diab0lic 11y ago
Whoops, I posted the mobile link from a gas station (traveling at the moment!). I apologize. Any chance a mod could swap the link? Thanks!
303.
by diab0lic 11y ago
I should note that the team responsible for this (Insight Engineering) is hiring a manager[0] and engineers[1]. If you're interested in solving this kind of problem come check out the job post as I'm really excited to be building
304.
by diab0lic 11y ago
Service owners wanted the system to be able to identify these problems as quickly as possible so the initial goal was to go as small as possible, however the initial data source rolled data up into 1 minute windows so we figured 5 minutes w
305.
by diab0lic 11y ago
One of the authors here, I'll be around to answer any questions if anyone has them. I'm sure my colleague will be around as well.
306.
Outlier Detection at Netflix(techblog.netflix.com)
274 points by diab0lic 11y ago|62 comments
307.
by diab0lic 11y ago
From the article: "At night, if there is no turbulence, Borschberg will be able activate the autopilot and nap, but only for 20 minutes at a time."
308.
by diab0lic 12y ago
Costless copies. I hear this one a lot, one of my last room mates used to say this one a lot. Its so black and white isn't it? How stupid am I that I don't understand this? Copies cost nothing, of course they don't. It doesn&
309.
by diab0lic 12y ago
The issue with naming conflicts arises from the fact that these things are generally created and named internally long before they're released as open source.
310.
by diab0lic 13y ago
One of the key words in the article's title was "secret" and further from the article "There will be absolutely no mention of United States government involvement,". If the intention were specifically to cause unres
311.
by diab0lic 13y ago
I unfortunately have this same problem, most of the time the IDE works great and it's feature set is most definitely nice. However sometimes it just locks right up / chugs along, especially with the IdeaVIM emulator -- which isn&#
312.
by diab0lic 13y ago
Is it though? If I made changes that broke something surely I would want to know everything that broke rather than just the first test in an arbitrarily ordered set of tests! Though I like your style, very simple, and simple is good. The st
313.
by diab0lic 13y ago
I was very eagerly awaiting the optimizations mentioned in part 1[0]. Unfortunately the optimizations work primarily on the Kepler architecture (Geforce 6xx, 7xx, and the GridGPU chips on EC2) and not the older Fermi chips that are so popul
314.
by diab0lic 13y ago
Is it really a witch hunt? People are just fascinated with the enigma that is Satoshi. He's (they've?) done something amazing in many people's eyes but has avoided taking credit for it -- in an arena where people are often se
315.
by diab0lic 13y ago
I'm sure they could at the very least attempt litigation for outright lying. Maybe false advertising? -- I suppose not since the FSF isn't selling anything in the traditional sense of the word. In any case that would be a PR night
316.
by diab0lic 13y ago
You're definitely right here, I don't think the pro line does anymore and the air never has. Small USB sticks aren't much more expensive, however. Especially if you get something as tiny as 256MB and just have a network based
317.
by diab0lic 13y ago
I can't imagine they were protesting the literal act of using a bus to ferry employees to work. Rather this was an easy way to target members of a group the protestors feel wronged by/animosity towards.
318.
by diab0lic 13y ago
Thats interesting, I'm from Yellowknife, Northwest Territories and I think everyone is on these superboxes and have been for a very long time. I didn't even realize delivery to the house was a thing until I moved south.
319.
by diab0lic 13y ago
Its been a long time since I wrote Ruby code, but if I recall correctly I believe the community is overwhelmingly in favour of 2 spaces as opposed to tabs.
320.
by diab0lic 13y ago
Females, at least where I'm from, regularly keep things in their bra such as cell phones. It never ceases to amaze me when a phone suddenly materializes from seemingly nowhere. Given that I feel they would hardly be opposed to some thi
321.
by diab0lic 13y ago
Possibly, though at this point after stumbling through using EC2 for these resources (and contending with the insane price fluctuations recently) I'm coming to the conclusion that owning my own hardware may be of substantial benefit fo
322.
by diab0lic 13y ago
I didn't realize the performance dip was so dramatic. As of late I've been considering acquiring some hardware and I guess I'm going to have to keep this in mind. If ones were to buy one today I don't even think I can fi
323.
by diab0lic 13y ago
If I had a 760 I'd likely just run all my experiments right on my device. As it stands my only machine is a Early 2009 macbook pro with a Geforce 9600M. The nice part about pylearn2, and theano is that the symbolic expressions compiled
324.
by diab0lic 13y ago
I have this same issue and am forced to run all my experiments on AWS. Once you get a workflow down it is pretty easy and the AWS resources provide an acceptable level of performance. Though I suppose if you're just toying around &#x2F
325.
by diab0lic 13y ago
If you execute the code shown in the README it should run an experiment against the MNIST dataset[0]. If it performs as well as other deep learning examples[1][2] you should see a test error rate of less than 1% on classification of handwri
326.
by diab0lic 13y ago
This is very interesting especially for those users more interested in running deep neural nets than programming deep neural nets. I was initially disinterested because it is missing several important features (at the moment) such as auto e
327.
Ask HN: Workflow for Scientific Computing in the Cloud?
9 points by diab0lic 13y ago|1 comments
328.
by diab0lic 13y ago
It may well be similar material to Ng's Coursera course, but the table of contents on the right shows that this is obviously going to transition into the topic of deep networks -- part of Ng's research but not his Coursera course.
329.
by diab0lic 13y ago
Hey Toby, First off thanks for writing the book. It was the first book recommended to me for ML while I was doing my undergrad and it got me using ML as opposed to studying ML. The book opened my eyes to the fact that I really could do
330.
by diab0lic 13y ago
Thats true, though as my optimization friends put it, sometimes a really close answer that is easy to come to is often better than a precise answer which is difficult to come to. Selecting the right algorithm is a bit of an art though some

More ›