mbq

6 ms·

1.
by mbq 10y ago
Um, are you sure we are there yet? For me it seems that the only atoms it has learned are the long-lasting static scenes, rather than eyes or mouths. Maybe it is just a matter of scoring and can be improved, but still...
2.
by mbq 11y ago
If it runs, it has to be decrypted (at a current level of cryptography); at most it is obfuscated and the access is blocked by some hardware tricks which may be costly to circumvent, but there is nothing fundamental stopping you.
3.
by mbq 11y ago
Binaries can be converted back to assembly and quite often even back to equivalent C; bugs are most often found by fuzzing (intentional or not) which does not require source code. The difference between open and closed source is that open i
4.
by mbq 11y ago
C is not a problem -- you can make a bug in every language. Even with memory safety and a perfect compiler, bug may direct the flow in bad direction (bypassing auth for instance) or leak information via side-channel.
5.
by mbq 11y ago
This is only for certain, verbal questions like antonym of, synonym of, X is to Y as Z to what; for each of them a separate, specific model was built (not that the AI read the question and printed relevant answer). To sum up, the paper shou
6.
by mbq 11y ago
About hard as setting up any UNIX daemon on a few computers. Reliability is a complex topic, depends on what you need -- for an anecdotal evidence I maintain a microscopic (10-20TB/4-6 desktop hosts) Moose deployment for a distributed
7.
by mbq 11y ago
There are also distributed file systems (MooseFS, Gluster, Ceph) which can duplicate data across physical locations, protect it with checksums, scrub in a distributed manner and auto-heal, all of that transparently.
8.
by mbq 11y ago
We do know what goes on in them. They are just trying (more-less, but it is only a matter of speed) random solutions till good enough, where the human operator decides what is "good". Which is fundamental, because when you have a
9.
by mbq 11y ago
This can be solved by containers.
10.
by mbq 11y ago
And Akamai's is inspired by http://http2.golang.org/gophertiles (which they acknowledge).
11.
by mbq 11y ago
I don't know the details, but this is rather due to fact how docker operates -- there is a daemon that runs with root privs (which are esential to create a container) controlled by a client with a protocol that has no concept of fine-g
12.
by mbq 11y ago
Why, docker-initiated chroot would prevent FF from acessing any not-FF+libs files, including which this malware steals; on Qubes it would have access to everything user-readable in the AppVM, which may include some secrets as Qubes workflow
13.
Should parsimony really still be the gold standard in modelling?(stats.stackexchange.com)
1 points by mbq 11y ago|0 comments
14.
by mbq 11y ago
Honestly speaking, "colour" cameras we use every day are also false colour; they shoot 3 monochrome images in red, blue and green chunks of spectrum and mix the whole thing using plethora of parameters like white balance, or gamma
15.
by mbq 11y ago
Isn't it actually written to the EFI system partition on disk rather than flashed into the MB?
16.
by mbq 11y ago
As with Java, the problem is not Flash per se but the browser plugin implementing SWF runtime. Flash can and probably will live long and prosper as the Flash "creator" app will switch to generating JavaScript+HTML5 istead of SWF.
17.
by mbq 11y ago
Still, I would read something titled "China's Market Crash and the Man Who Earned >10M$ Having Precisely Predicted It".
18.
by mbq 11y ago
It shouldn't be significant because GW is all about how bad Earth is in dissipating energy into space (in comparison to black body), rather than how much energy flows through. Nevertheless, let's count: according to Wiki, 2012 Wor
19.
by mbq 11y ago
Also pulling anything from CDN basically means that CDN operator (or anyone that will manage to hack it) can spy on or alter communication between your users and your server.
20.
by mbq 11y ago
Even with PZ DNA would have major and minor groove rather than being a symmetrical double helix beloved by virtually all illustrators, sadly also those of pop sci articles...
21.
by mbq 11y ago
98% of computer-using population will use whatever comes with the machine; Linux is viable but the hardware vendors are not pushing it, also because of Microsoft lobbying. On the other hand Android could be easily built on top of QNX and ge
22.
by mbq 11y ago
Nope; *ply are implemented as loops, you are only saving time on reallocating memory structures. Real speed-up is gained from using vectorised functions and operators (implemented in C or Fortran in R engine or a certain package) for whole
23.
by mbq 11y ago
The execution could be better, but it is certainly not a stupid idea. GA code is esentially a concealed, unconditional transmit of the user data to a third party, and allowing this third party to intercept and modify your communication; pre
24.
by mbq 11y ago
The fact that they want beacons suggests that the location information is not as accuracte or reliable as they would like to; quantitative changes also matter.
25.
by mbq 11y ago
FB knows the list of visitors from their query logs and in principle may later sell it to the business either directly or as a more targeted ad. Still this beacon network may be also alternatively mapped (or ripped from FB servers) by the c
26.
by mbq 11y ago
Turboprop business is an another airline, eurolot; it changes, but it is often a separate company.
27.
by mbq 11y ago
I bet the author calls for making functions just right in size to leverage self-containment and complexity, instead of panicking after crossing some magical number of lines. This is actually compatible with DRY, because overpartitioned func
28.
by mbq 11y ago
Still, this also confirms the obvious fact that the selection of the winning team based on that hidden, private test also gives no guarentee that it is not accidently overfitted, especially with tons of submissions. The only reliable option
29.
by mbq 11y ago
It is a common misconception and a huge source of disappointment with ML -- without proper validation of the whole model building procedure (method selection + parameter tuning + feature selection + fitting) no amount of data and magic tri
30.
by mbq 11y ago
It is sad that the post fails to acknowledge that hyperparameter tuning may also be a source of overfitting (and very often is) -- one should always treat is as a part of training and validate as such, for instance by using nested CV. And n

More ›