Jump to content

Wikipedia:Reference desk/Archives/Computing/2021 January 20

From Wikipedia, the free encyclopedia
Computing desk
< January 19 << Dec | January | Feb >> Current desk >
Welcome to the Wikipedia Computing Reference Desk Archives
The page you are currently viewing is a transcluded archive page. While you can leave answers for any questions shown below, please ask new questions on one of the current reference desk pages.


January 20

[edit]

Wireless Mouse

[edit]
  • If someone uses an wireless mouse, and uses the built in keyboard on a laptop, can the hacker hijack the keystrokes using the wireless mouse receiver? --つがる Talk to つがる:) 🍁 01:12, 20 January 2021 (UTC)[reply]
    It seems highly unlikely. I am not sure what you mean by "hijack the keystrokes"; there could be exfiltration of what keys I am typing, or spoofing of keys that I did not type, things like that, in theory, but it would depend completely on the kind of hardware in play. There are different types of laptops and different types of wireless mice and endless permutations of them together.
    Let me take a common case as an example. The user has an ordinary laptop and a Logitech wireless mouse with a unifying receiver. Now, the unifying receiver is a special case. It does not use PINs or security to sync devices. It is also a general-purpose input device controller which can have more than one slave. So if the attacker is close enough!!! and manages to sync to the unifying receiver as a keyboard, then the attacker could conceivably execute keystrokes onto the laptop while I'm using the wireless mouse. The attacker would not be able to see the things I am typing on the built-in keyboard, and should not be able to see the mouse movements or button presses.
    Now if we switch the scenario to Bluetooth, that should make it somewhat resistant. If the user is careful, then no unknown devices can pair, under normal circumstances, and so theoretically the wireless mouse could be used without fear of hijack. But Bluetooth has insecurities, so if the attacker were able to pair as an input device, the attacker could be a keyboard or a mouse or both or anything she wants. And it would not be necessary for me to be using a wireless mouse, just having Bluetooth enabled would be enough.
    But all this is highly unlikely in most cases because of two things, proximity and vigilance. Your attacker will need to sidle up very close to your laptop. Sitting next to you at Starbucks. And you would need to be unaware that your keyboard is going haywire. Most people would notice right away. But then the damage could be done? Elizium23 (talk) 05:06, 20 January 2021 (UTC)[reply]

Free reminder software

[edit]

I'm looking for a program where I can make a note that I need to do something on X date, and the program reminds me when I log into my PC on that date. Suggestions? Clarityfiend (talk) 08:29, 20 January 2021 (UTC)[reply]

Cannot Windows calendar application do this? Ruslik_Zero 10:55, 20 January 2021 (UTC)[reply]
Mozilla Thunderbird and its forks have calendar functionality. Alternatively if what you need to do is on the PC, you can skip the reminder phase and set it to automatically start/open on that date via Windows Task Scheduler. 93.136.149.88 (talk) 23:27, 21 January 2021 (UTC)[reply]
So, like, what OS? Elizium23 (talk) 06:23, 22 January 2021 (UTC)[reply]
Windows 10. Clarityfiend (talk) 07:34, 23 January 2021 (UTC)[reply]

Machine Learning Algorithm Tutorial

[edit]

Could anyone point me toward a written or maybe video tutorial on how to construct a machine learning algorithm from scratch. Specifically I'm interested in decision tree and neural network. I watched a great tutorial on YouTube that really did cover a lot, but it basically just explained how to implement existing python libraries. I don't want to reinvent the wheel, but I do want to have as comprehensive understanding of how these algorithms actually work. TheRiseOfSkittlez (talk) 21:31, 20 January 2021 (UTC)[reply]

Personally, I'd start by reading an implementation of least squares regression in the language of your choice.
If you can't see the connection to a higher parameter-space, re-read our article on regression analysis: almost every machine learning problem is representable as a regression problem over a generalized set of variables. In the case of a neural network, the parameters you are fitting are the numbers - the numerical coefficients - that describe the connectivity of the "neural" nodes. In the case of a decision tree, the parameters you are fitting are the branch probabilities.
If you follow LSQR, you can probably conceptualize the generalization to other fitting methods; actually implementing those methods is pretty painful, but here's a zero-cost full-length textbook that guides the reader through the underpinning math.
Nimur (talk) 22:34, 20 January 2021 (UTC)[reply]

The video course at https://fast.ai is very good. 2601:648:8202:96B0:0:0:0:313A (talk) 06:21, 22 January 2021 (UTC)[reply]