Wikipedia:Reference desk/Computing

Welcome to the computing section
of the Wikipedia reference desk.

skip to bottom

Select a section:

• Computing
• Entertainment
• Humanities
• Language
• Mathematics
• Science
• Miscellaneous
• Archives

Want a faster answer?

Main page: Help searching Wikipedia

   

How can I get my question answered?

• Select the section of the desk that best fits the general topic of your question (see the navigation column to the right).
• Post your question to only one section, providing a short header that gives the topic of your question.
• Type '~~~~' (that is, four tilde characters) at the end – this signs and dates your contribution so we know who wrote what and when.
• Don't post personal contact information – it will be removed. Any answers will be provided here.
• Please be as specific as possible, and include all relevant context – the usefulness of answers may depend on the context.
• Note:
  • We don't answer (and may remove) questions that require medical diagnosis or legal advice.
  • We don't answer requests for opinions, predictions or debate.
  • We don't do your homework for you, though we'll help you past the stuck point.
  • We don't conduct original research or provide a free source of ideas, but we'll help you find information you need.

Ready? Ask a new question!

How do I answer a question?

Main page: Wikipedia:Reference desk/Guidelines

• The best answers address the question directly, and back up facts with wikilinks and links to sources. Do not edit others' comments and do not give any medical or legal advice.

See also:

• Teahouse
• Help desk
• Village pump
• Help manual

June 30

PAC learning - the function of the probability distribution D

Hi,

I am trying to understand the basic definition of PAC learning from Shai Shalev-Shwartz's "understanding machine learning". They define a hypothesis class to be PAC learnable if for every distribution D over the instances, and for any labeling function f, an approximately correct hypothesis can be learned with high probability over the random choice of a training set. Two issues that are not entirely clear to me:

(1) what is exactly the function of D? isn't the requirement of "for every D" very strict? What if D samples unrepresentative training set? for example, in a digit recognition task, what prevents D from sampling only the digit 6 - how could an hypothesis selected based on that training set have a small generalization loss?

(2) They define PAC learnability as a property of an hypothesis class, i.e., of a solution. Intuitively I'd expect that learnability would be a property of the problem, as some problems are harder than others. What role do the properties of the problem play in the definition?

Thanks! — Preceding unsigned comment added by 87.71.54.101 (talk) 18:17, 30 June 2018 (UTC)

Wikipedia articles: PAC (Probably approximately correct learning), Error tolerance (PAC learning).

Answers in Quora: [1]

Video lectures: PAC Learning, [2]

D is the fixed unknown probability distribution from which some instances x are drawn for training. Intuitively I'd define PAC learnability as demonstration that a particular hypothesis strategy applied to a given problem D provides results that converge to ever decreasing error as more instances are learned. Early geocentric modelling of planet orbits that required arbitrary hypotheses of epicyclic motions was overthrown by the more PAC-learnable heliocentric model that was refined in accuracy by Newton (hypothesis of orbital motions) and Kepler (source of observation data). DroneB (talk) 12:23, 1 July 2018 (UTC)

July 1

FACEBOOK - NO SOUND

There is no sound when i play Facebook videos. How can I overcome this problem? I use Windows 10. Thank you123.231.124.13 (talk) 03:02, 1 July 2018 (UTC)

There's usually a volume control button in the bottom right hand corner that usually needs to be clicked to turn on the sound. If that's not working, make sure that other programs have sound. If they do not, then the sound for the computer is turned down or off. If they do, then the video on Facebook may not have sound. Ian.thomson (talk) 03:05, 1 July 2018 (UTC)

Odd network setup

I have a router (Vodafone Connect, if it matters) that delivers a signal to most of my property. Using powerline ethernet (which works well), I have connected the LAN port of a second router (because I don't have a dedicated switch) to the first router. I did not connect to the WAN port of the second router because I am aware of the problems of trying to use multiple shared resources through NAT.

Unfortunately, the second router has but one LAN and one WAN port. I was hoping, with a little ingenuity, to connect a VOIP adaptor to the WAN port of the second router such that I can use it as I would like. The VOIP adaptor (presumably) only needs a handful of services on the internet, and nothing specifically on the LAN.

But, doing so naively does not result in a working configuration for my VOIP adaptor! I'm not sure what to do. I've tried port forwarding, but I'm not sure precisely what it is, let alone how to use it effectively. Any ideas?--Leon (talk) 19:26, 1 July 2018 (UTC)

You should connect the second router (to the first router) using its WAN port and then switch it to the bridge mode. Ruslik_Zero 20:41, 1 July 2018 (UTC)

Doesn't have it! The WAN settings it has that look useful are static IP and DHCP. I can set routes and port forwarding on it, but I've no idea what to do.--Leon (talk) 21:56, 1 July 2018 (UTC)

What is the model of the second router? Ruslik_Zero 18:42, 2 July 2018 (UTC)

A Pentagram P6360.--Leon (talk) 05:36, 4 July 2018 (UTC)

Have the 2nd router no WAN IP configurtation and no PPPoE login. Have it just be a client device, getting an IP address as a DHCP client as computers are configured as "automatically" by default. Ensure, the 3rd byte of the IPv4 address differs on each of Your LAN networks.

If this does not fulfill Your need of NAT access and You do not want to route NAT twice, have the 2nd router this other configuration: Change the fixed IP address of the 2nd router into an avail IP address of the 1st routers LAN. The 1st router is still connecting to Your ISP and does not need any change! In the same step disable the DHCP service of the 2nd router. If possible, have the change at once before restarting the 2nd router with Your config. When the IP address is set, connect the LAN of the 1st router to the LAN 2nd router. If there is not 1GBT (=1000MBIT) LAN or auto-negociate functionality of the PHY, You may need a Ethernet crossover cable if there is no connection when cable is connected. When the routers are the same brand, You may create a temporary entry in the file hosts when administrating the 2nd router again which is not necessary if the 2nd router does not reroute to a domain name in its administration site. The 2nd router now acts as a dedicated WLAN accesspoint and ethernet hub/switch, whatever the device is designed. --Hans Haase (有问题吗) 10:40, 6 July 2018 (UTC)

A 3rd way is to configure the provider login data into the 2nd router and degrade the 1st to a DSL modem, only if these modes are supported by the routers. --Hans Haase (有问题吗) 10:44, 6 July 2018 (UTC)

July 2

Improving unintelligible speech

How can I extract the words of a hardy intelligible audio file? I tried reducing the speed, and cleaning it with the noise removal feature of Audacity. Anything else that can be done? --Doroletho (talk) 02:17, 2 July 2018 (UTC)

There's been a lot of work recently in using AI to extract a voice from a noisy recording. It may be you can find one on the web which will try your recording and give you the result. I await the first instance of one that is fooled into saying something completely different the way AI recognition of pictures has been ;-) Dmcq (talk) 13:03, 2 July 2018 (UTC)

Extracting voice from all the rest would be a real improvement. I wonder what makes voice different from everything else. But indeed, no matter in what language it is, humans seem to have no problem differentiating between what's speech and what's noise.--Doroletho (talk) 15:04, 2 July 2018 (UTC)

Humans sometimes have difficulty differentiating between speech and noise; according to DL Murman (Predicting When Dementia Starts, 2015[3]) “speech comprehension in the setting of background noise and ambiguous speech content declines with age….” The OPs aim is speech comprehension and not necessarily speaker recognition so some non-linear processes such as clipping and/or companding that distort the sound may prove beneficial. Here are research papers [4], [5] and a discussion at StackExchange of human speech noise filtering. DroneB (talk) 14:39, 3 July 2018 (UTC)

I find that closing my eyes while listening improves intelligibility a heck of a lot, without having to do anything to the recording. 173.228.123.166 (talk) 05:04, 5 July 2018 (UTC)

I didn't thought about that, but it makes sense. Maybe reducing the visual processing makes our brains concentrate more on audio input.

I also tried reducing the sound volume with mixed results. This might work because you also reduce the noise, and you force yourself to pay attention to what you are listening to. Doroletho (talk) 10:20, 5 July 2018 (UTC)

Reciprocally, I sometimes find that noise makes reading more difficult. —Tamfang (talk) 02:35, 6 July 2018 (UTC)

Remote desktop

I work for a large state agency. We have hundreds, perhaps thousands computers. They are all managed by another state agency. The way my work is structured I do have some gaps which sometimes last an hour or more. I have some software projects home and on the weekends I now debug a large C# application. I won't mind to set up a remote desktop home to be able to do some work from here, at work, on my home project.

I am not sure I can change the OS to do remote desktop at work, but I may try. A small window is available for this at work, and surely at home.

I wonder if doing this will be detectable by the technical staff who monitor our computers? Thanks, AboutFace 22 (talk) 16:02, 2 July 2018 (UTC)

Yes, of course technical staff can simply monitor all connections using their network. Jahoe (talk) 09:21, 3 July 2018 (UTC)

• Your employer's IT is able to monitor all inbound and outbound data packets from your computer to the internet (assuming a non-headbangingly-stupid configuration of the internal network). Cryptography may obfuscate the contents of such communications, but basic metadata will always be visible: for instance in HTTP(S) the network needs to know to which IP address and port to deliver the packets. Hence it will always be "detectable" in the sense that IT can see stuff going up and down the wires.

What IT can deduce from the information they see depends on a lot of complex factors, including what exact data they can observe in the clear, what amount of data goes through in which pattern, what are other typical uses, etc. A desktop application will probably look extremely fishy on traffic charts compared to normal internet usage (steady traffic instead of spikes of page loads). Tigraan^{Click here to contact me} 14:01, 3 July 2018 (UTC)

@Tigraan and other fellow, thank you. I won't contact you, @Tigraan. I've got what I needed. In a small way I am already there. I don't have a remote desktop, but I often connect to the stuff on the web having nothing to do with my job description. It is MSDN forums, C# and C++, SQL Server, a lot of math of various kind. I do numerical integration. So far nothing happened. I guess I won't do the remote desktop, I will survive without it. Our Internet access is blocked to some places I occasionally run into by accident, but I don't need them. Nobody restricts the stuff I am interested in, so it must be OK. Thank you again, AboutFace 22 (talk) 15:08, 3 July 2018 (UTC)

You should bear in mind that your employer almost certainly has an Internet Usage Policy or similar. My employer's policy says that "occassional" use of the internet at work for private purposes is acceptable provided it does not adversely impact your work or the network performance. There are a long list of "unauthorised" activities such as downloading software, playing online games etc. Probably a good idea to check your employer's equivalent policy, rather than assuming that anything that is not blocked is fine. Gandalf61 (talk) 13:54, 5 July 2018 (UTC)

@Gandalf61, thank you. I vaguely remember reading something like this. The problem is that my employer absolutely cannot structurally occupy me full time. I wish they could. My nature is to do something and what I do here is interesting, but I do everything fast and efficient. Then I wait for emergencies. I cannot install any software because even when I get a message that Adobe needs an update and I click YES, the update is rejected. They do it all centrally. I expected there might be a problem in implementing remote desktop but I got authoritative information that it would be better for me to stay out of it. Thank you all. AboutFace 22 (talk) 16:36, 5 July 2018 (UTC)

July 3

Removing Facebook app from Android

After a decade of using iOS, I've recently changed to Android. I can't seem to uninstall Facebook App, rather only disable it. Seemingly it's a built-in app (wtf?), but didn't I have to download it first? I'd like to remove the app as it's too data-consuming. Please help me to uninstall it on the Samsung J7. 31.217.26.62 (talk) 00:29, 3 July 2018 (UTC)

Some Samsung devices have Facebook app pre-installed as a system app. So, it can not be uninstalled, only disabled. Ruslik_Zero 20:34, 3 July 2018 (UTC)

Wikipedia preview on Google and Bing

When one searches on both Google and Bing, one is presented with a preview of the Wikipedia article at top right of the results page. I have noticed an odd difference between the two in the case of David Lloyd George. Google calls him "Former British Prime Minister", Bing "Former Chancellor of the Exchequer". Now, both are true, but it is as PM that he is more famous, and he was PM after being Chancellor. Do we know why the two search engines give different short descriptions? DuncanHill (talk) 13:47, 3 July 2018 (UTC)

It is a wrong place you posted at. You should go to Wikipedia:Help. They will take care of it. Alternatively you can make the change yourself. AboutFace 22 (talk) 15:12, 3 July 2018 (UTC)

Err no - Wikipedia Help is not the place to explain why two external search engines give different results. AS for "make the change myself", I don't have that level of control over Bing. DuncanHill (talk) 15:15, 3 July 2018 (UTC)

It's not a preview of the Wikipedia article. Only a text paragraph ending with "Wikipedia" or "en.wikipedia.org" is quoted from Wikipedia. We often get blamed for errors in the other parts and made {{HD/GKG}} and {{HD/Bing}} for the purpose, as well as {{HD/YKG}}. PrimeHunter (talk) 19:31, 3 July 2018 (UTC)

I'm not blaming us for their errors. I'm asking how they come up with the short descriptions and why they differ. DuncanHill (talk) 20:50, 3 July 2018 (UTC)

Did you read {{HD/GKG}}? They are not Wikipedia:Short description and are not taken from Wikipedia but from an unspecified source or a combination of search results. They are different companies with their own algorithms and I see no reason to expect them to come up with the same string. You could try a guess by Googling "David Lloyd George" "Former Chancellor of the Exchequer". Many results describe him like that, e.g. a 2017 book called David Lloyd George: Former Chancellor of the Exchequer. You could also try looking at the html source. Google says: <div class="sthby kno-fb-ctx"><span data-ved="0ahUKEwim_8uO-oPcAhVP_KQKHWkBDdAQ2koIiAIoAjAa">Former British Prime Minister</span></div>. I don't know whether this contains a hint of where they got it. Bing says: <div class="b_entitySubTitle">Former Chancellor of the Exchequer</div>. That doesn't leave room for a potential source hint. PrimeHunter (talk) 22:28, 3 July 2018 (UTC)

getting circled Arabic digits to display properly

I typed up a passage in Arabic, with circled digits, and at first they displayed properly. Then, one day, the digits appeared outside the circles. It doesn't matter if I change fonts (I was using TNR), and I get the same result when I post on WP as I do in LibreOffice. However, if I double the circle, the digit(s) will appear in one of them:

۝٧۝

Well, sometimes they will. Even with copy&paste, sometimes the digit(s) will appear between the circles. (Right now, both here in the edit window and in page preview, I see the 8 in the left-side circle above.) I assume the problem has something to do w forcing LTR/RTL, with the change in direction of Arabic numbers compared to letters, but I have no idea how.

Oh, in this

٧۝

orientation, the digit is reduced in size, as if it were in the circle, even though it appears to the left of it.

Hm, in the search window of my browser (FF), it displays properly. I suppose it could be an update to TNR that introduced a bug, but according to the MS website for TNR, there haven't been any updates.

Any ideas? — kwami (talk) 19:06, 3 July 2018 (UTC)

I wouldn't even know how to start to get what you're doing and I know a fair deal about computers. I suspect that editors on the Language reference desk might be more help. Dmcq (talk) 12:17, 6 July 2018 (UTC)