0534: "Genetic Algorithms"

This forum is for the individual discussion thread that goes with each new comic.

Moderators: Magistrates, Prelates, Moderators General

0534: "Genetic Algorithms"

Postby Ctta0s » Fri Jan 23, 2009 5:07 am UTC

Image
Title text: Just make sure you don't have it maximize instead of minimize.
Link: http://xkcd.com/534/

Am I the only one confused by this? Maybe teh Wiki will answer my questions.

P.S. Why no thread?
EDIT: Until now, obviously. That was like 8 minutes or something.
Last edited by Ctta0s on Fri Jan 23, 2009 5:09 am UTC, edited 1 time in total.
Ctta0s
 
Posts: 31
Joined: Thu Jul 17, 2008 7:25 am UTC
Location: Various

Re: "Genetic Algorithms" Discussion

Postby Steve the Pocket » Fri Jan 23, 2009 5:08 am UTC

cephalopod9 wrote:Only on Xkcd can you start a topic involving Hitler and people spend the better part of half a dozen pages arguing about the quality of Operating Systems.

Baige.
User avatar
Steve the Pocket
 
Posts: 458
Joined: Mon Apr 23, 2007 4:02 am UTC
Location: Going downtuuu in a Luleelurah!

Re: "Genetic Algorithms" Discussion

Postby joee » Fri Jan 23, 2009 5:09 am UTC

hahahahaha i love you randall.

Ctta0s: read up on genetic algorithms?
Hi glasnt.
User avatar
joee
 
Posts: 227
Joined: Mon Dec 03, 2007 5:53 am UTC

Re: "Genetic Algorithms" Discussion

Postby DarthMarth » Fri Jan 23, 2009 5:09 am UTC

I almost just made this my first comic thread.

I think I'll appreciate this one more in a year or two once I learn what genetic algorithms are.
User avatar
DarthMarth
 
Posts: 16
Joined: Mon Aug 11, 2008 3:07 pm UTC

Re: "Genetic Algorithms" Discussion

Postby TheHand » Fri Jan 23, 2009 5:09 am UTC

Well there's a thread now! Yeah, I was confused too, what can I say. But I'm not complaining! XKCD didn't get where it is today by dumbing it down. So if that means I have to do some research then so be it!
User avatar
TheHand
 
Posts: 178
Joined: Mon May 05, 2008 8:03 am UTC

Re: "Genetic Algorithms" Discussion

Postby kaniman2 » Fri Jan 23, 2009 5:09 am UTC

I'm confused too.

If people explain it to me, I promise to laugh. Even if jokes aren't as funny when you explain them at all. Now I really wish I paid attention when my mom was trying to teach me some programming languages.

And holy crap, four replies above already.

EDIT: Oh, lol.
Last edited by kaniman2 on Fri Jan 23, 2009 5:10 am UTC, edited 1 time in total.
Hiya, I'm Kanavazk.
[(No longer) bored out of my mind...]
Spoiler:
[Well, it's better than Vista. --Meowgan]
(I DO realize that that quote was from an xkcd comic, thank you very much.)
User avatar
kaniman2
 
Posts: 284
Joined: Fri Jan 23, 2009 3:18 am UTC
Location: Canada, North of the Border, whatever you want to call it.

Re: "Genetic Algorithms" Discussion

Postby aion7 » Fri Jan 23, 2009 5:10 am UTC

Remember that line. All it takes is one person forgetting...
Spoiler:
Zeroignite wrote:And you have suddenly become awesome.

joshz wrote:Oh, you so win.

internets++ for aion7.

jerdak wrote:Nothing says hello like a coconut traveling near the speed of light.
User avatar
aion7
 
Posts: 1142
Joined: Mon Dec 03, 2007 1:43 am UTC
Location: In a base with which you identify, killing dudes to whose team you belong

Re: "Genetic Algorithms" Discussion

Postby mollusk » Fri Jan 23, 2009 5:11 am UTC

Might I recommend a WillingnessToOpenPodBayDoors constant?
John Hodgman wrote:...while the truth may be stranger than fiction, it is never as strange as lies . . . or as true.
User avatar
mollusk
 
Posts: 33
Joined: Mon Aug 11, 2008 4:55 pm UTC

Re: "Genetic Algorithms" Discussion

Postby aion7 » Fri Jan 23, 2009 5:12 am UTC

Make sure you keep the variable neurotoxins in a private method far away from any AI.
Spoiler:
Zeroignite wrote:And you have suddenly become awesome.

joshz wrote:Oh, you so win.

internets++ for aion7.

jerdak wrote:Nothing says hello like a coconut traveling near the speed of light.
User avatar
aion7
 
Posts: 1142
Joined: Mon Dec 03, 2007 1:43 am UTC
Location: In a base with which you identify, killing dudes to whose team you belong

Re: "Genetic Algorithms" Discussion

Postby Ctta0s » Fri Jan 23, 2009 5:12 am UTC

I've had an algorithms course, just didn't cover genetic algorithms. They seem... interesting (from the 10 seconds of Wiki reading I did). And I got the Skynet reference, really. I'm an avid follower of Summer Glau... errr, Terminator: The Sarah Connor Chronicles.
Ctta0s
 
Posts: 31
Joined: Thu Jul 17, 2008 7:25 am UTC
Location: Various

Re: "Genetic Algorithms" Discussion

Postby madhollywood » Fri Jan 23, 2009 5:14 am UTC

Wow, I am glad that I am not the only one who was a bit confused. I got the terminator reference too and even know a little about genetic algorithms... I guess I am amused by my own confusion.
User avatar
madhollywood
 
Posts: 69
Joined: Thu May 15, 2008 12:13 am UTC

Re: "Genetic Algorithms" Discussion

Postby niteice » Fri Jan 23, 2009 5:14 am UTC

Genetic algorithms are an AI topic, that's why.
GENERATION 4294967292: The first time you see this, copy it into your sig on any forum, negate the generation, and convert it to a 32-bit unsigned integer. Social experiment.
User avatar
niteice
 
Posts: 186
Joined: Wed May 02, 2007 4:17 am UTC

Re: "Genetic Algorithms" Discussion

Postby PCChris » Fri Jan 23, 2009 5:15 am UTC

Basically, this fitness function determines what breeds to reproduce. Higher cost = less likely to reproduce and breed with each other. Therefore, if the code is going to become skynet, then it shouldn't breed.

http://en.wikipedia.org/wiki/Fitness_function
PCChris
 
Posts: 5
Joined: Fri Jan 23, 2009 5:12 am UTC

Re: "Genetic Algorithms" Discussion

Postby Ctta0s » Fri Jan 23, 2009 5:17 am UTC

No, it's ok to have the neurotoxins variable, just make keep the morality core away from incinerators. ;)
Ctta0s
 
Posts: 31
Joined: Thu Jul 17, 2008 7:25 am UTC
Location: Various

Re: "Genetic Algorithms" Discussion

Postby brainfreeze215 » Fri Jan 23, 2009 5:17 am UTC

Great. Now I'll have to wait until Monday for a comic I understand.
brainfreeze215
 
Posts: 5
Joined: Wed Feb 06, 2008 3:50 pm UTC

Re: "Genetic Algorithms" Discussion

Postby VHBT » Fri Jan 23, 2009 5:21 am UTC

mollusk wrote:Might I recommend a failureToOpenPodBayDoorsCost? [edited to fit the format better]

...and a lyingAboutCakeCost?

Also, think about the implications of the alt text. The difference between Math.max() and Math.min() is only two characters, and if even one person in the world got it wrong...
VHBT
 
Posts: 50
Joined: Fri Mar 28, 2008 8:18 pm UTC

Re: "Genetic Algorithms" Discussion

Postby dennisw » Fri Jan 23, 2009 5:22 am UTC

The implication is that genetic algorithms could lead to self-aware computers, a la Skynet from the Terminator films (see reference in an earlier post). This is also in keeping with the idea of a technological singularity in which machine intelligence surpasses human intelligence and the potential problems resulting such as the decision by the machines that we are inferior and unnecessary.

See also Asimov's Three Laws of Robotics.
Try the Printifier for xkcd. You can now scale the comic between 50 and 150%.

I find these very useful: Common Errors in English Usage (web site) and Eats, Shoots & Leaves (book). You may, too.

e pluribus unum
Unleash unlicensed ungulates!
User avatar
dennisw
 
Posts: 438
Joined: Wed Nov 05, 2008 9:09 am UTC
Location: Appearing pro se AND pro bono!

Re: "Genetic Algorithms" Discussion

Postby Eternal Density » Fri Jan 23, 2009 5:23 am UTC

Wow, I'm been reading a book about population genetics and yesterday I was talking about it with a guy (a newly started colleague) who did a lot of GA stuff at university (more than I did, which was only part of a semester, sadly)... so now it's my turn to say "Get out of my head, Randall!"
The colleague was explaining how he made a multi-layered GA so that the genetic algorithm itself is optimised by a genetic algorithm. This leads to some interesting results. It also caused me to mentally paraphrase, "Yo dawg, I heard you like genetic algorithms, so I put a genetic algorithm in your genetic algorithm so you can evolve while you evolve."
Then I started daydreaming about an AI which learns how to most effectively learn, and what that would mean in a practical (i.e. writing fiction about it) sense.

Closer to the real topic, all efforts as described in this comic are wasted if you type 'import skynet'.
Although, if Skynet is sitting in a module on your computer ready to be imported, you're already doing it wrong...

Oh, and we need to include failureToProtectCompanionCubeCost too.
ImagePlay the game of Time! castle.chirpingmustard.com Hotdog Vending Supplier

GAZEBO OINTMENT HARPSICHORD CREDENZA BUNGALOW
User avatar
Eternal Density
 
Posts: 3820
Joined: Thu Oct 02, 2008 12:37 am UTC
Location: The Hotdog Cart

Re: "Genetic Algorithms" Discussion

Postby dwibby » Fri Jan 23, 2009 5:25 am UTC

VHBT wrote:Also, think about the implications of the alt text. The difference between Math.max() and Math.min() is only two characters, and if even one person in the world got it wrong...


Eh, it's not that bad. You spend the better part of a day running from a homicidal robot or two, and the problem basically takes care of itself. Just make sure you know where your local metalworks is and you'll be fine.
dwibby
 
Posts: 3
Joined: Mon Oct 27, 2008 1:31 pm UTC

Re: "Genetic Algorithms" Discussion

Postby videogamesizzle » Fri Jan 23, 2009 5:26 am UTC

hey hey! ah ha ha! ha! ha...ha...hu...heh.

seriously, i don't get it. i gotta pay more attention. and watch the terminator movies.
Look at me still talking when there's SCIENCE to do!
Silvyr wrote:I fucking love cocaine. I wish I could buy it somewhere...
User avatar
videogamesizzle
 
Posts: 358
Joined: Sat Oct 25, 2008 3:32 am UTC
Location: Rockford, IL

Re: "Genetic Algorithms" Discussion

Postby diotimajsh » Fri Jan 23, 2009 5:30 am UTC

Ohhh, okay, I see. (I hadn't known what Skynet was until referring to this thread.)

The fact that it's written in Python, I believe, is a subtle reference to The 2008 Christmas Special.



...Friggin' Python.
Osha wrote:Foolish Patriarchy! Your feeble attempts at social pressure have no effect on my invincible awesomeness! Bwahahahaa


Blog type thing
User avatar
diotimajsh
 
Posts: 658
Joined: Wed Nov 14, 2007 7:10 am UTC

Re: "Genetic Algorithms" Discussion

Postby dennisw » Fri Jan 23, 2009 5:31 am UTC

PCChris wrote:Basically, this fitness function determines what breeds to reproduce. Higher cost = less likely to reproduce and breed with each other. Therefore, if the code is going to become skynet, then it shouldn't breed.

http://en.wikipedia.org/wiki/Fitness_function


Yeah

http://en.wikipedia.org/wiki/Fitness_function wrote:Interactive genetic algorithms address this difficulty by outsourcing evaluation to external agents (normally humans).


"normally" = "for the time being" :shock:
Try the Printifier for xkcd. You can now scale the comic between 50 and 150%.

I find these very useful: Common Errors in English Usage (web site) and Eats, Shoots & Leaves (book). You may, too.

e pluribus unum
Unleash unlicensed ungulates!
User avatar
dennisw
 
Posts: 438
Joined: Wed Nov 05, 2008 9:09 am UTC
Location: Appearing pro se AND pro bono!

Re: "Genetic Algorithms" Discussion

Postby Seraph_Kreigs » Fri Jan 23, 2009 5:36 am UTC

[First time post, long time reader]

I registered just to make a comment about this post.

I'm actually writing a genetic algorithm right now in MATLAB for a NASA Aeronautics contest... and I about died laughing when I read today's comic. In fact, I will find a way in my fitness function to add this line, even if it's commented out. =)
Seraph_Kreigs
 
Posts: 1
Joined: Fri Jan 23, 2009 5:32 am UTC

Re: "Genetic Algorithms" Discussion

Postby Iluvatar » Fri Jan 23, 2009 5:39 am UTC

For those who don't want to click any of the reference links, a quick explanation of Genetic Algorithms:
1) You have a problem
2) You generate a bunch of things that look like solutions
3) You run each solution through a 'fitness function' that determines how 'good' that solution is.
4) Generate a bunch of new solutions based on the most fit of the previous generation
5) Repeat 3 + 4 until you get a really good solution.

So, if your fitness function makes a solution that include 'become skynet' be really unfit (have a high cost), the algorithm will not propagate those solutions.

Guess this was a good semester to take Intro to AI. :) Next week we learn about neural networks, though, and I suspect they'll be a lot more Skynet prone.
"Don't you see?!"

"Get out of my bed, Randall, I'm trying to sleep!"

Your results may vary.
User avatar
Iluvatar
 
Posts: 143
Joined: Wed Jan 24, 2007 2:19 pm UTC

Re: "Genetic Algorithms" Discussion

Postby Unforgiven » Fri Jan 23, 2009 5:45 am UTC

Imagine the carnage if someone accidentally goes overboard, and sets the value too high for its data type so that it wraps around to a negative number.

Instant SkyNet! :)
"Now we're at the museum. Do you think they'll have DDR in there too?"
"It's Japan. Of course they will."
User avatar
Unforgiven
 
Posts: 162
Joined: Wed Jul 25, 2007 8:48 am UTC
Location: Tokyo, Japan

Re: "Genetic Algorithms" Discussion

Postby phlip » Fri Jan 23, 2009 5:48 am UTC

Unforgiven wrote:Imagine the carnage if someone accidentally goes overboard, and sets the value too high for its data type so that it wraps around to a negative number.

Instant SkyNet! :)

Well, this is Python... bignums are standard.
While no one overhear you quickly tell me not cow cow.
but how about watch phone?
User avatar
phlip
Restorer of Worlds
 
Posts: 7162
Joined: Sat Sep 23, 2006 3:56 am UTC
Location: Australia

Re: "Genetic Algorithms" Discussion

Postby scwizard » Fri Jan 23, 2009 5:59 am UTC

This type AI research seems very complex, difficult and time consuming. I'm surprised that it's had the modest success that it's had.

EDIT: I'm not saying here "computers and maths are hard", what I'm saying is that this is the study of heuristics using heuristic methods in the area of computer science where heuristics don't work as well as they do in other fields.
~= scwizard =~
scwizard
 
Posts: 519
Joined: Sun Mar 04, 2007 6:29 pm UTC
Location: New York City

Re: "Genetic Algorithms" Discussion

Postby sargeras0000 » Fri Jan 23, 2009 6:09 am UTC

Meh. Seeing as I'd never heard of genetic algorithms until now, I feel proud that I got it, if after a few seconds pondering. Now, OFF TO THE GOOGLE!
sargeras0000
 
Posts: 41
Joined: Tue Oct 28, 2008 4:05 am UTC

Re: "Genetic Algorithms" Discussion

Postby Beautiful_Solace » Fri Jan 23, 2009 6:12 am UTC

Sooooo confused Algor...algorithms?!??!?! lol is that what we're talking about j/k but seriously i don't get it :/
"Mother is God in the eyes of a child"
-Silent Hill
User avatar
Beautiful_Solace
 
Posts: 12
Joined: Fri Dec 05, 2008 2:59 am UTC
Location: Southern Ontario, Canada

Re: "Genetic Algorithms" Discussion

Postby Beautiful_Solace » Fri Jan 23, 2009 6:13 am UTC

Beautiful_Solace wrote:Sooooo confused Algor...algorithms?!??!?! ¡This cheese is burning me! is that what we're talking about j/k but seriously i don't get it :/

who put the cheese thing in there?
"Mother is God in the eyes of a child"
-Silent Hill
User avatar
Beautiful_Solace
 
Posts: 12
Joined: Fri Dec 05, 2008 2:59 am UTC
Location: Southern Ontario, Canada

Re: "Genetic Algorithms" Discussion

Postby Garnett » Fri Jan 23, 2009 6:16 am UTC

Code: Select all
PROBE BEHAVIORS (With assigned priorities).
Communicate (5).
Record Data (4).
Analyze Data (3).
Seek Replication Materials (999).
Move to Current Target (1).
Garnett
 
Posts: 7
Joined: Sun Aug 10, 2008 10:32 pm UTC

Re: "Genetic Algorithms" Discussion

Postby dennisw » Fri Jan 23, 2009 6:38 am UTC

Beautiful_Solace wrote:
Beautiful_Solace wrote:Sooooo confused Algor...algorithms?!??!?! ¡This cheese is burning me! is that what we're talking about j/k but seriously i don't get it :/

who put the cheese thing in there?

Skynet
Try the Printifier for xkcd. You can now scale the comic between 50 and 150%.

I find these very useful: Common Errors in English Usage (web site) and Eats, Shoots & Leaves (book). You may, too.

e pluribus unum
Unleash unlicensed ungulates!
User avatar
dennisw
 
Posts: 438
Joined: Wed Nov 05, 2008 9:09 am UTC
Location: Appearing pro se AND pro bono!

Re: "Genetic Algorithms" Discussion

Postby PhantomPhanatic » Fri Jan 23, 2009 6:40 am UTC

Nice one Randall.

Really though, doesn't the risk of having insta-Skynet when accidentally maxing cost sort of outweigh the benefits of averting Skynet-like species from reproducing? What is the probability that Skynet-like species will develop compared to the probability that a programmer might accidentally max cost?
We can lick gravity, but sometimes the paperwork is overwhelming.
-- Wernher Von Braun
User avatar
PhantomPhanatic
 
Posts: 84
Joined: Thu Nov 13, 2008 5:32 am UTC

Re: "Genetic Algorithms" Discussion

Postby of.hippo » Fri Jan 23, 2009 7:19 am UTC

Hilarious! Practically the xkcd paradigm comic: genius.

Don't ever quit, xkcd. Promise?
of.hippo
 
Posts: 10
Joined: Mon Oct 29, 2007 10:05 pm UTC

Re: "Genetic Algorithms" Discussion

Postby Bunsen » Fri Jan 23, 2009 7:20 am UTC

Garnett wrote:
Code: Select all
PROBE BEHAVIORS (With assigned priorities).
Communicate (5).
Record Data (4).
Analyze Data (3).
Seek Replication Materials (999).
Move to Current Target (1).


I came here just for this.

(Oh! My! That was a bit extreme, wasn't it?)
Bunsen
 
Posts: 4
Joined: Tue Apr 01, 2008 2:41 pm UTC

Re: "Genetic Algorithms" Discussion

Postby honnza » Fri Jan 23, 2009 7:34 am UTC

Garnett wrote:
Code: Select all
PROBE BEHAVIORS (With assigned priorities).
Communicate (5).
Record Data (4).
Analyze Data (3).
Seek Replication Materials (999).
Move to Current Target (1).

what if it does the higher priority things first? I guess the world will be quite surprised soon.
honnza
 
Posts: 7
Joined: Tue Jan 06, 2009 12:50 pm UTC

Re: "Genetic Algorithms" Discussion

Postby plams » Fri Jan 23, 2009 7:37 am UTC

I came here just for this.

(Oh! My! That was a bit extreme, wasn't it?)


Will you believe I got a mid-air collision with you because I posted the same god damn thing?! Star Control 2 reprazzenT!
plams
 
Posts: 35
Joined: Fri Oct 13, 2006 12:08 am UTC
Location: denmark

Re: "Genetic Algorithms" Discussion

Postby Nemphael » Fri Jan 23, 2009 7:59 am UTC

Meh, managed to post in the other thread, me.

thisPostBecomingSkynet = 2^99DISREGARD THAT I SUCK COCK

GAH, I'll have to watch Terminator sometime. I've just watched the I think... third one? Oh, well, back to calculating GPS positions.
User avatar
Nemphael
 
Posts: 422
Joined: Fri Sep 12, 2008 9:44 pm UTC

Re: "Genetic Algorithms" Discussion

Postby Gero » Fri Jan 23, 2009 8:00 am UTC

Great idea Rhandal. Very important public service anouncement.

mollusk wrote:Might I recommend a WillingnessToOpenPodBayDoors constant?


Also good idea, along with the weighted companion cube preservation ones.

Now if only we can find an algorithm to protect us from lightsabre-wielding raptors....



brainfreeze215 wrote:Great. Now I'll have to wait until Monday for a comic I understand.


Sorry dude. This one is really worth renting Terminator and reading up on wikipedia for. It's hilarious when you understand it. really.
Gero
 
Posts: 58
Joined: Fri Jan 02, 2009 1:29 pm UTC

Re: "Genetic Algorithms" Discussion

Postby Amarantha » Fri Jan 23, 2009 8:06 am UTC

diotimajsh wrote:The fact that it's written in Python, I believe, is a subtle reference to The 2008 Christmas Special.
Or a blatant reference to Python :P
omgryebread wrote:...if I'm watching a fantasy, I want my princess to stab some motherfuckers, claim the crown herself, then invade the prince's kingdom and sleep with his sister.
User avatar
Amarantha
 
Posts: 1639
Joined: Tue Nov 27, 2007 4:56 am UTC
Location: Melbourne, Australia

Next

Return to Individual XKCD Comic Threads

Who is online

Users browsing this forum: 4ax235vwt7, brenok, Google [Bot], mrob27, Neil_Boekend, PayasYouDraw, PM 2Ring, rmsgrey, Sustainabilizer, TheMinim and 15 guests