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

December 17

Need help with Python and a Firefox plugin to download YT playlists

So I've come upon this helpful thread[1] on how to download entire playlists from YouTube. However, all this Python coding and programming BS is just beyond me. What does the guy mean by "adding" VLC and youtube-dl, and how do I do this? I've barely managed to install Python and youtube-dl (though I'm not even sure if I've even managed to install the latter), and when I use the "add" button, I can only paste something into two boxes which looks like a.) custom program name, and b.) required program path.

However no matter what I do, I don't seem to get VLC and youtube-dl "added", all I've managed is to make self-titled entries show in the addon options page that seem to be pointing nowhere and where (unlike in that guy's screenshot of how it's supposed to look) not even a logo shows up next to the self-titled name that I gave it. I can now right-click on any link and go "Open with...youtube-dl" or "Open with...VLC", but neither of them does anything at all. --46.93.158.170 (talk) 11:32, 17 December 2018 (UTC)

Hm. I seem to have at least made VLC available to the plugin now, but when I go to "Open with...VLC" like the linked thread is telling me to, no download of the entire playlist happens, it only opens one single video in VLC from cache. --46.93.158.170 (talk) 11:37, 17 December 2018 (UTC)

Dammit, after downloading YouTube DL GUI (which, of course, only worked as a GUI to download single videos, not playlists) that I came upon by mere chance while googling desperately for youtube-dl's documentary, I actually had to write my own Python programming script. Of course, *NOTHING WHATSOEVER* works how the youtube-dl documentation[2] claims it does *AT ALL*. *NOT A SINGLE ONE* of those "official" commands and options seem to even be recognized by youtube-dl, nor does the bare-bones download of youtube.dl *WITHOUT A GUI* from Github even just provide a program that can be used in command prompt, only lots of stupid "py" files that Windows can't even seem to be able to use or recognize. Only the third-party GUI comes with an actually executable .exe that it runs from cache, as it shows you in the program options via a file path to a temporary folder, after you've hit a "DOWNLOAD .exe" button in the GUI. So I could at least execute this temporary cache .exe via command prompt while the GUI was open, whereby I found out that it recognizes none of the "official commands" at all.

So, no way to tell youtube-dl to simply extract the audio from the videos and only download those or specifically which items I want from the YouTube playlist, not even any way to specifiy a download folder other than the contemporary cache folder, but *DAMN*, the Python programming script I finally came up with during a course of maybe 2 hours of trying one script after another did at least download the entire playlist, and the programming script looks like this: [TEMPORARY CACHE]\youtube-dl.exe [PLAYLIST URL]. Phew!

In any case, that entire thread I've linked in my OP above is entirely misleading. You don't need any plugin for Firefox, all you need is to install Python and then YouTube DL GUI which will temporarily download you an .exe that you can use in command prompt. Even though it recognizes none of the commands and options from the official youtube-dl documentation, it can at least be used via command prompt to download entire YouTube playlists in one go. --46.93.158.170 (talk) 12:47, 17 December 2018 (UTC)

I have no comment on your other problems but youtube-dl does explicitly provide a command line Windows executable that comes with Python and should work without requiring any additional dependencies. I used it myself about 3 weeks ago.

This is the link [3]. It is from this page [4] where it is described thusly "We also provide a Windows executable that includes Python." where Windows executable is the link.

It is also listed here [5] "Latest (v2018.12.17) downloads: youtube-dl youtube-dl.exe youtube-dl-2018.12.17.tar.gz See the right for more resources." and here [6] "Windows users can download an .exe file and place it in any location on their PATH except for %SYSTEMROOT%\System32 (e.g. do not put in C:\Windows\System32)." so shouldn't be hard to find.

Note that when I used youtube-dl ~3 weeks ago, several command line options did seem to work as expected. As said, I did use the Windows executable provided by someone in the dev team and not random versions from random GUIs or whatever so that could be the reason.

I actually ended up using streamlink [7] not youtube-dl in the end since I couldn't work out or maybe it wasn't possible [8] [9] to download an ongoing live stream from the ~earliest possible. Although I can't explicitly remember if I tried the -live-restart option [10] but do remember there was some weirdness with the hls options I did try although it seemed to be a Youtube compatibility issue and not because youtube-dl wasn't working as expected. (Not strictly the beginning/start since it was Al Jazeera so it's one of those continuous live streams that goes back about 4 hours at any one time.) But I was able to use youtube-dl to download from the current position.

Nil Einne (talk) 03:12, 18 December 2018 (UTC)

Alphabet Ink corporate office phone number

I need to contact someone at the Corporate Headquarters of Alphabet Inc but I’m having difficulty finding a phone number or email. Does anyone have that information? I have called directory assistance and news papers in Mountain View, CA and searched the Internet but have been unsuccessful. — Preceding unsigned comment added by 184.103.117.30 (talk) 17:31, 17 December 2018 (UTC)

We have an article Alphabet Inc. that gives the headquarters location as Googleplex, 1600 Amphitheatre Parkway, Mountain View, California, U.S. and their website that invites investor contact at an email address given at the bottom at this link. DroneB (talk) 17:54, 17 December 2018 (UTC)

December 19

"Please format SD card to FAT 32 before using..." ???

I am thinking of buying this dash cam. Tucked away on the page is the instruction: "Please format SD card to FAT 32 before using..." What does this mean and how do I do it? P.S. Any real advantage to buying a higher memory card for this--a 128 or 256 instead of a 64? Longer record time? Better reesolution video? Thanks--67.244.114.239 (talk) 06:41, 19 December 2018 (UTC)

Anything that stores data, needs a way to find this data back. There exist various variants of such systems, but FAT32 is one of the oldest and most common ones. The device itself can likely format the SD card, it will be in the instructions. Please be aware that when you format an SD card, you will loose any information on it. Newly bought SD cards often come preformatted as FAT32 (but not always).

The advantage of a bigger card is that it allows you to store more video. A dashcam will override itself once it fills up, but this size determines the age of the oldest video that it will be able to keep. For this setup, a 64GB SD card will hold about 3,5 hours of video and a 128GB about 7 hours. More important however with this particular camera is to pay attention to the speed of the card as they are actually 2 cameras which requires a high speed card. You need something that is 4K capable (often called a U3 card or a V30 or up). You will also want a proper brand, like Sandisk or Samsung. I personally use a "Samsung EVO Plus Micro SDXC 64 GB" for similar purposes and it has not failed me yet. —TheDJ (talk • contribs) 08:32, 19 December 2018 (UTC)

If you are using Windows or Mac, the best program for reformatting an SD card is the official one from the SD Association. You can find it at [ https://www./downloads/formatter_4/ ]. --Guy Macon (talk) 10:55, 19 December 2018 (UTC)

Thanks all. Got it. And doubling the overwriting time for just a few bucks more feels prudent, so going for the 128.--67.244.114.239 (talk) 14:42, 19 December 2018 (UTC)

Something mangled the link. Here it is. --47.146.63.87 (talk) 21:48, 20 December 2018 (UTC)

Can my ISP personnel see which pages of a website i visited

I browse internet by first visiting a private IP address 172.16.0.1 and then logging in with username allotted to me by my ISP and a password.I want to know if my ISP can see exactly what pages of a website like http://en.wikipedia.org I have visited.Another thing is from ethernet port at the back of my CPU cabinet a CAT 5/CAT 6 cable comes out and goes to wifi router box at my building top with PoE power supply that is provided by my ISP.The box modem is of BATON make as far as I know.Please help.103.24.110.150 (talk) 07:34, 19 December 2018 (UTC)

As long as you use HTTPS (secure web) version of websites, which most are these days, then your ISP can only see which domains (websites) you have visited, not the pages that you have visited (within it). —TheDJ (talk • contribs) 08:14, 19 December 2018 (UTC)

It's a little-known fact, but thanks to the WMF it is possible in some cases for third parties to tell exactly what Wikipedia pages you read, even though you used https. The technical details are complex, but they are explained at Wikipedia:Village pump (policy)/RfC: Wikimedia referrer policy.

Consider the case where a Wikipedia user reads our page on Bomb-making instructions on the internet and then clicks on the link to Feinstein Amendment SP419 at Cornell university. A government agency then gets a court order giving them access to Cornell university's server logs (or someone simply hacks the server to get the logs). If we were a silent referrer (which the WMF rejected because it makes their requsts for donations less effective), the logs at cornell.edu would simply show that a particular person read https://www.law.cornell.edu/uscode/text/18/842, a perfectly innocent act. Because we send domain-only referrer information, the logs at cornell.edu will say that that person clicked on a link to the Feinstein Amendment SP419 page while reading Wikipedia, and a link search using the tool we helpfully provide will show that the only link to www.law.cornell.edu/uscode/text/18/842 on Wikipedia is from our Bomb-making instructions on the internet page. So, by sending referrer information, we just turned the government knowing that a particular user accessed the text of the Feinstein Amendment SP419 -- a perfectly innocent act in itself – to the government knowing that a particular user accessed the text of the Feinstein Amendment SP419 while reading the Wikipedia Bomb-making instructions on the internet page. Don't forget the couple who were questioned by the police after he did a google search on "backpacks" while she did a Google search on "pressure cookers" in another room... --Guy Macon (talk) 11:11, 19 December 2018 (UTC)

This is true of Web pages in general. Even if you visit an HTTPS site, by default, if your browser loads anything on other sites because the page tells it to (scripts, ads, etc.), it will send that site the referer, so now that site knows what you visited. And, if that site isn't HTTPS, anyone can read that going over the wire. Also modern browsers can, with prefetching, follow regular links without you even clicking on them. The Web was never designed to be a secure communications medium, and most Web browsers prioritize convenience and speed over security. If you really want to hide what you're visiting from a savvy attacker, you need to take a lot of precautions and have some computing know-how. Web searches will turn up various guides. --47.146.63.87 (talk) 21:40, 19 December 2018 (UTC)

Not true. In general (the user can override the default behavior), browsers only send referrer information if the website doing the referring tells them to.

Depending on how we configure Wikipedia if someone loads an external image or clicks on an external leak in our Bomb-making instructions on the internet page, the external site could get any of the following in the referrer:

• en.wikipedia.org/wiki/Bomb-making_instructions_on_the_internet#[Section] (full URL)
• en.wikipedia.org/wiki/Bomb-making_instructions_on_the_internet (domain plus page)
• en.wikipedia.org (full domain)
• wikipedia.org (partial domain)
• No referrer information (silent referrer)

--Guy Macon (talk) 22:35, 19 December 2018 (UTC)

I admit I'm not an expert Web dev, but I was under the impression that browsers by default send a referer whenever they fetch a resource linked in a tag. I just checked this by loading nbc.com on stock Firefox, and every resource fetch passed a referer header. A site can instruct the browser to behave differently, either through a rel tag or JavaScript, but this is deviating from the usual behavior (and of course requires browser support). Vis a vis the original question, you should assume your browser is sending referer headers, unless you've configured it not to. Of course, a bigger problem is JavaScript, which can do basically anything it wants. --47.146.63.87 (talk) 21:45, 20 December 2018 (UTC)

Another issue is that the "referer" might be acting secretly like part of the URL. Like, it at least used to be that if you went to a New York Times article directly they would give you some paywall crap, but if you copypasted the title into Google, ran the search, and clicked on the result they would display the article from the "same" URL. Really, if the browsers were on our side they would have the referer in/near the address bar. Wnt (talk) 15:54, 21 December 2018 (UTC)

December 20

Samsung Session Expired Tone

I have a Samsung Galaxy J7, which, if I am correct, is running Android version 8.1.0. My question is that, sometimes, after I have been using it, and then have not been using it for a while, it pings me with an audible tone and says Samsung Session Expired. My question is: How do I get it to stop giving me an audible tone to say that my session expired? I don't really care what a Samsung session is, but I don't want to have to pick up the phone and look at it to see if I just got a text message when all that it is doing is telling me nothing. Can I turn these stupid tones off? Robert McClenon (talk) 03:25, 20 December 2018 (UTC)

The Internet has two solutions listed (I haven't tested these because I don't have a Samsung phone):

Keeping your Samsung account:

• Tap Settings.
• Scroll down and tap Samsung Account
• press Cancel Sync
• Reboot
• Check your notifications and re-enter password if needed.

Nuking your Samsung account:

• Open Settings
• Select Accounts and Sync
• Choose Samsung
• Select your samsung account email
• Tap remove account

--Guy Macon (talk) 16:42, 20 December 2018 (UTC)

Thank you. I will try the first. Nuking the account isn't worth it. Robert McClenon (talk) 18:40, 20 December 2018 (UTC)

Minor differences. Samsung Account is under Accounts. Cancel Sync is done by turning off Auto Sync (and it then tells me that turning off auto sync saves some battery power but requires manual sync for data, but I am not trying to keep data synchronized between my desktop Dell, my laptop Dell, and my smartphone. I don't want to live in a social media utopia, because a utopia is literally a no-place. Robert McClenon (talk) 17:39, 21 December 2018 (UTC)

Samsung Not Giving Audible Tone

Is there a condition that sometimes prevents me from getting an audible tone when I receive a text message? (This is almost the opposite annoyance of the one mentioned above.) Or is this simply due to my having the phone in my pocket or being in a noisy place or having my car radio on loudly? Robert McClenon (talk) 03:25, 20 December 2018 (UTC)

As a test, you could send a message to yourself, with a 1 minute delay (scheduled message). Do it several times, with the phone in and out of your pocket. However, this doesn't help if you've assigned custom message tones for certain people. LongHairedFop (talk) 11:50, 21 December 2018 (UTC)

"Amoeba" "solves" NP-hard problem in "linear time"

This is really cool, but I think it's all a trick: [11][12]. They have an "amoeba" (really, a true slime mold, Physarum polycephalum), which comes up with an approximate solution to the traveling salesman problem in apparently linear time. The amoeba has help from a computing process that looks at whether its "solutions" to the problem have inconsistencies: it has a channel for which numerical order each city is visited in, so say B1 to B5 for whether city B is visited first or fifth, and if any number or letter is visited by one pseudopod all the conflicting channels get hit with light to make others retract.

My assumption is that because they claim to have a decent simulation for the amoeba, which works in "almost linear time", using only N^2 parallel processors I assume, then they can't possibly get to beating exponential time unless they have solved the NP-hard problem in computer science by having an "amoeba" show them how to code. So I would suppose that they are modelling an exponential using a polynomial for some limited N and then eventually something will go drastically wrong, even though they say they want to get a bigger photolithography apparatus to make thousands of channels for the plasmodium to travel into.

Question is, am I right about this? It's not really within my competence, and if they have a magic tool to break our encryptions I ought to find out. Wnt (talk) 15:35, 20 December 2018 (UTC)

As you have described it, it isn't necessarily a trick or a breakthrough. I haven't read the paper, but you say that it comes up with an approximate solution in linear time. That isn't theoretically interesting. The crux of the problem has never had to do with approximate solutions. What is NP-complete is finding an exact solution. The amoeba is probably just a device for implementing a greedy algorithm, and greedy algorithms usually work very well on the traveling salesman problem. They just don't always work, and occasionally they get stuck. A greedy algorithm is used in road navigation systems, and usually works very well, but doesn't always come up with the best answer, and once in a while the navigation system can send you in a circle around the destination. There isn't anything special about approximate solutions. The problem is that finding an exact solution is NP-hard. Robert McClenon (talk) 16:44, 20 December 2018 (UTC)

(edit conflict)They say the time grows linearly as the number of cities increases from four to eight. I can do that with ease: just build in some sort of inefficiency that makes the four-city case take longer.

But can it be done without such trivial cheating? See Why My Soap Film is Better than Your Hadoop Cluster

Nature solves such physical systems automatically, for instance the n-body problem. There is a physical way to sort in linear time: cut a stick to the length of each of the items to be sorted. Bang one end of them on the desk, then start picking the tallest one, repeat. Bubba73 ^{You talkin' to me?} 04:11, 21 December 2018 (UTC)

BTW, as far as I know, no widely-used cryptography system is based upon traveling salesman. Now if they found a faster way to solve elliptic curve or pairs of large primes, that could break some popular algorithms. --Guy Macon (talk) 17:08, 20 December 2018 (UTC)

Having read the Wikipedia article, I see that this observation is very interesting biologically. It seems that the amoeba or slime mold has some sort of memory at a cellular level that is not in the nervous system that it does not have. This may provide insight into the evolution of the nervous system, for instance. However, it isn't interesting mathematically. The significance to crypto is that crypto would be one of the many interesting results of solving NP-complete problems. But this isn't a solution to an NP-complete problem, since the traveling salesman problem isn't to find an approximate solution, but an exact solution. An exact solution to traveling salesman would be a solution to every NP-complete problem. Would be. Robert McClenon (talk) 18:39, 20 December 2018 (UTC)

The nervous system angle was what attracted me to the article to begin with, but in this particular case, it sounds like there is nothing very sophisticated going on. The pseudopods are retracted in response to light, contributing cytoplasm to a central storage, and the cytoplasm evenly expands everywhere until light shines on it again. There was some commentary about synchronized waves of expansion (I think it was) but their simulation didn't need to implement them, and AFAICT they just meant more pressure throughout the whole organism at once. Wnt (talk) 15:50, 21 December 2018 (UTC)

December 21

memory management and a large hash table

I'm using what is basically a large hash table. I'm trying to get as much data as possible into a Windows computer with 128GB of RAM. I'm using Delphi. The number of rows (the hash values) is fixed but the number of items for each hash varies from 0 to over 65000 (and is not known in advance).

My first approach was to use a list of lists. But when I added a new item to the list, it seemed to double the amount of memory allocated, which got to taking up too much memory.

I switched to a dynamic array of dynamic arrays, with the idea of when I needed more data for a particular hash, I'd increase the corresponding array by a certain percentage (I tried 10%, 20%, and 25%). But initially it seems to be using a lot more memory than expected. In the test I'm running now, (has been running for a few hours and will finish overnight), the memory use according to task manager got up to 97%, 98%, and then 99%. It is constantly adding data as it runs. But after a while, the memory use dropped to 83%, then to 70%, and now it is at 62% (even though it is rapidly adding data).

So what I'm wondering is, when I increase the size of a dynamic array by a few percent, is it actually allocating double that (the way it does with lists), and then as it runs, the system is smart enough to realize that it isn't using a lot of this memory and releasing it? Bubba73 ^{You talkin' to me?} 04:04, 21 December 2018 (UTC)

Now it is down to using only 55%, even though it adds about 200,000 8-byte pieces of data per second. You might think that I could let it use more memory to begin with and then let it scale back, but that causes it to swap to HD, which brings it to a crawl. Bubba73 ^{You talkin' to me?} 05:51, 21 December 2018 (UTC)

To answer such a question about how the internal memory of the dynamic array are managed, you will need to tell us exactly what version of Delphi compiler and runtime support libraries you are using.

For example, if you're using Embarcadero Delphi, the official product documentation wiki describes the platform memory manager. On Windows, the 'FastMM' system backs your arrays, and it uses a block-based allocation algorithm, rounding the amount of memory up to the nearest block size.

Modern memory management algorithms are complicated by confounding details of the internal implementation. I would posit that the actual implementation details so complicated that it would be prohibitively difficult for us to give you a very detailed answer. You could empirically measure and benchmark the performance, or you could get in touch with your software vendor if it's really important; or you could write your memory-sensitive code in a language that lets you get closer to the operating-system and hardware memory details, without putting a friendly runtime support library between you and the actual memory layout.

Nimur (talk) 17:30, 21 December 2018 (UTC)

I'm using the almost-current version 10.2.3 (on 64-bit Windows 100. The program finished a few minutes ago. When I got up today, it was back to using 99% of the memory, but it didn't seem to be swapping to HD (at least I didn't hear any of the constant thrashing to the HD). Bubba73 ^{You talkin' to me?} 18:51, 21 December 2018 (UTC)

It says "For Win32 and Win64, the Memory Manager employs an algorithm that anticipates future block reallocations, reducing the performance impact usually associated with such operations. The reallocation algorithm also helps reduce address space fragmentation." So that must be why it seems to double the allocated space for both lists and dynamic arrays. But I didn't see anything saying that it releases unused memory. The Task Manager reported that consistant dropping of memory usage even as the memory was being filled up. Bubba73 ^{You talkin' to me?} 19:00, 21 December 2018 (UTC)

Diagram of a symmetric multiprocessing system

Rent time on Google Compute Engine or one of its competitors?

Buy a used 256GB server for a couple of thousand bucks? https://www.amazon.com/gp/offer-listing/B075XR4G18/

--Guy Macon (talk) 00:09, 22 December 2018 (UTC)

I'd like to do that. I bought a used HP Z420 workstation with 128GB last month. Bubba73 ^{You talkin' to me?} 00:15, 22 December 2018 (UTC)

Question: with a dual CPU system like that, will each CPU have access to all of the memory or does each of the two chips get half of it? Bubba73 ^{You talkin' to me?} 00:18, 22 December 2018 (UTC)

Each CPU access all of RAM. They each have their own cache so as to minimize waiting for memory. --Guy Macon (talk) 16:16, 22 December 2018 (UTC)

Thanks, then it looks like I have to have more than one CPU chip to get more memory slots. And a related question: I have 16GB sticks of DDR3 ECC in my HP Z420, even though the docs say a maximum of 8GB sticks (which were probably the maximum at the time). Are there sticks with more capacity that would work? (I found 32GB sticks, but I'm not sure if they are right. The documentation says DDR3 ECC or non-ECC, unbuffered.) Bubba73 ^{You talkin' to me?} 03:14, 23 December 2018 (UTC)

December 22

Array in VBA

Hello! I'm learning macro in Excel, but VB's array syntax is difficult for me. Can anyone rewrite the follow Lua code in VBScript?

-- No built-in reverse method in Lua, thus create a function
local tableReverse(tab)
    local tabRev = []
    for i, v in ipairs(tab) do
        table.insert(tabRev, 1, v)
    end
    return tabRev
end

local myArray = ["A1", "A2", "A3", "A4", "B1", "B2"] -- It's annoyed that we have to set a length in VB, count items manually is boring

table.insert(myArray, "B3")  -- VB doesn't have push method, again...

for i = 1 to 5 do -- Push five items
    table.insert(myArray, "C" .. i)
end

for i, v in ipairs(myArray) do -- For Each ... Next in VB?
    print(v)
end

local myArrayReverse = tableReverse(myArray)

for i, v in ipairs(myArrayReverse) do
    print(v)
end

--아리랑&아리랑 (talk) 13:47, 22 December 2018 (UTC)

December 23

Super-URLs with HTTP headers?

An obvious deficiency in URLs is that they don't contain all the data to get to the desired page. For example, a New York Times page would often provide a useless result unless (I assume) it had the right referer string saying it came from a Google search. Also User-Agent and so forth so readers don't get dinged over a browser if they follow a link, etc. More generally, you can't properly point someone else to exactly what happened for debugging without all the information. Maybe even HTTP methods could be specified, though I'm not quite sure what would happen. ;) It occurs to me that this is the sort of deficiency that someone might have envisioned plugging in the glory days of the Web, before 9/11/01. Did anyone ever establish some kind of super-url format standard with a syntax for listing any and all of the HTTP headers? Alternatively, it should make a fine browser plug-in. Anyone see such a thing? Wnt (talk) 02:53, 23 December 2018 (UTC)

December 24

Is machine learning == artificial intelligence?

Is all AI, at least, all practical applications of AI, nowadays just some flavor of ML? --92.191.143.129 (talk) 12:56, 24 December 2018 (UTC)

Artificial intelligence is a very broad field. Machine learning is a subset of artificial intelligence. 216.59.42.36 (talk) 14:25, 24 December 2018 (UTC)

Same OP, another IP here.

I'm slightly confused still about the whole picture surrounding AI, ML and things like deep learning, neuronal networks.

According to the box on the right of Artificial intelligence, machine learning is a major goal, but not an approach. As approaches it lists: symbolic, deep learning, bayesian networks, evolutionary algorithms.

On the other hand, Outline of artificial intelligence lists machine learning as just one approach. That is, it divides the approaches in symbolic (called Good Old Fashioned AI), subsymbolic (with a load of subdivisions, including ML) and statistical.

And why isn't ML classified into statistical? Isn't it somehow a way of generating stats about what to do in context or after X happens?

Why don't we have a rule-based approach in the AI article or outline? Wouldn't that be feasible and a complete different animal than the approaches mentioned above? --83.39.57.239 (talk) 17:09, 24 December 2018 (UTC)