| |
| |
| |
|
Page: 1 2
Comments:
<0> hello people
<1> hey
<2> hey bitches
<0> hi ulti
<0> do you have website ulti...
<1> do you mean do I have random links to information, or do I have a personal website?
<0> yah, personal website i mean?
<1> yeah I have http://www.mattoates.co.uk which just points to http://users.aber.ac.uk/meo3
<1> vanity URL's make me feel dirty :[ but they look pretty on CV's
<1> is charles_a related to christine_c ????
<0> no, i dont know him
<0> why? was he a regular member here
<1> no idea
<1> no one talks on here enough for me to know who is regular
<2> hehe, there are a few regulars
<2> Ceran, DanF_DrC , cwenner --
<2> what are yall interested in ?
<3> this channel is worthless
<1> t35t0r it could be good if 64 people started to talk ;)
<2> t35t0r, what was deep blue programmed on
<3> programmed on or in
<3> it's an ibm asc
<0> heheheh, ulti....can we integrate AI in spell checker?
<2> on , arhcitecture
<1> qbert alife specifically looking at evolutionary methods for developing game playing (at the moment)
<0> can you give some idea? is this also an NLP
<2> NLP is for losers, we're mostly Machine learning here
<3> nlp is for losers?
<3> ****y ou!
<0> i see
<1> lol
<0> sorry for that
<3> machine learning is lame
<0> im also a loser
<1> yeah NLP is a mugs game! ;)
<1> its too hard :'[
<3> irc is the best place for developing an nlp
<1> whereas machine learning could lead us to a modal that could /learn/ language :P
<1> t35t0r I imagine everyone here is keeping logs just for that, I know I am
<3> #gentoo
<4> worthless? losers? what is this? :)
<0> ulti...im looking now in your website, it seems that you are an expert on this
<0> hi cwenner
<4> #ai is logged by clog
<3> cwenner, that's what i said
<1> charles_a lol no
<1> I wouldnt use the #ai logs :/ its all just connect messages and no conversation
<1> erk christine_c even
<1> someone make an AI tab complete KTHNX
<1> I tried making a shell that understood my tab complete hammering, it didnt work so well
<3> did you just do a m*** invite?
<1> nope
<1> that would be awful netiquette
<1> and shameless plugging
<3> Ulti, use screen ..it goes woof woof
<3> when you tab hammer
<1> heh
<1> what I wanted to do was reduce all typing in the shell to mashing <tab> until the computer did what I was thinking
<1> using the shell normally is pretty much like that anyway
<5> hi everyone
<3> what i want to do is make an agent that i can say "find me directions to blah from my house and give me a printout"
<5> Would anyone happen to know what domain this game falls into, and if it's been solved? http://googlemodules.com/module/121/
<3> it solves into combinatorial mathematics and dividing a state space
<3> stupid
<3> i hate games
<3> i don't understand that game
<5> I don't either, but now that I'm trying to solve it, I'm getting addicted :)
<3> that's why i hate it
<5> The annoying part is I'm starting to get pretty good at it, but I don't know why.
<1> heh
<1> isnt that game purely down to chance?
<1> you just work out systematically which pegs to keep and which ones to change each time until you have the right combination
<5> Ulti: It's a little more than chance. Based on your guesses, you can narrow down the selections until you find the right colors.
<1> yeah
<5> The trick is making informative guesses. You'll never win by with just random trial and error.
<1> but the initial state is pure chance
<1> then you can limit it
<1> first you try the all the same colour which it gives you as default to work out how many are that colour :/
<1> its a boring game
<1> its just maths
<1> well I guess all games are just maths
<1> but it's maths you can do in your head systematically
<5> The strategy I use is to start guessing one color for all pegs. That'll either tell you there are no pegs with that color, in which case you can eliminate that color entirely from your future predictions, or it'll tell you exactly how many pegs have that color. Then you can start changing one peg at a time to narrow down the choices.
<5> Ulti: You sound like you haven't played it yet :)
<1> Cerin I just played it
<1> it was boring, I went back to playing bomber man
<1> :>
<5> heh, I guess it is boring if you can't win :P
<1> its about as fun as "guess the country I'm thinking about" which I played for 6 hours with some friends in the pub, except that was fun because you could narrow it down from give aways in their body language
<1> kind of like how poker is only fun if you bid, the rest is just probability
<1> we ran out of real countries and started making names up, amazingly we still managed to guess some ^_________^
<0> hi cerin, im still new to ai games, when you had design that, did you apply minmax strategy
<5> christine_c: design what?
<1> they should change the Turing test to see if a machine can guess arbitrarily made up country names from a human opponent
<0> your game, the one that you talk about
<5> christine_c: I didn't design that. I'm just trying to write a program to solve it.
<0> ok
<5> minimax wouldn't work because I'm not playing against an opponent
<0> ok, thanks for the info
<5> Ulti: If that game's so boring for you, you should have no problem writing a program to solve it.
<1> sure but doing one that always gives the minimum number of moves is a little harder
<1> there is 1296 possible combinations???
<1> you could make a graph that represents the whole search space and then update the whole thing with weightings on the edges for which colours are more common in the result you are looking for... so you first guess all 6 colours on their own
<1> but that is going to be slow and fairly space complex
<5> 4 pegs can be 1 of 6 possible colors, for 6*6*6*6 = 46656 possible combinations
<5> opps, no, you're right 6*6*6*6 = 1296
<5> used the wrong operator
<1> if there was a couple more pegs I would do it another way, its all just probability and search I hate that... too much of AI is just search (if not all)
<1> I figure one day all computers will just come with petabytes of random noise that we search for information
<6> Hi all, im new to AI and im wondering if a production rule system could feasibly be used in a chatbot to help it "learn" about family. I'm thinking rules could easily describe relationships to other family members...any thoughts?
<7> why a production rule system?
<6> erm probably only because i've heard of it and understand the concept
<7> i think they don't lend themselves well to learning environments. you should read some more about knowledge representation.
<7> some software that i know of that you might get started with is sneps or loom.
<6> ok, i'll take a look
<6> as this is a university project i'll need to write my own code rather than use existing systems though
<7> they're written in common lisp by the way.
<1> howcome everyone else is only just starting their universtiy projects and I've been doing mine for half a year :[ my uni ****s
<6> this isn't a project, like a final year project...its a single module
<7> maybe they're in a different hemisphere from you.
<1> doofy you might want to look at prolog family relationships is a well documented example program
<1> doofy http://www.amzi.com/manuals/samples/prolog/misc_one_file/gene.pro
<7> Ulti: you're quite right about prolog but... how do you go about writing a program that learns in just prolog?
<1> you could write a grammar parser to look for names and family labels, that ***erts those as instances in the family rules
<1> but yeah it wouldn't be trivial
<1> for a single module it's probably a bit much
<7> i've been trying to grasp logical resolution and deductive database systems, so please excuse my current tendency to look down my nose at prolog. ;)
<1> I've only written one prolog program :] to diagnose problems in a simplified chemical plant, and that was only for half of a modules credit
<7> i found an inspiring example of "extreme prolog" programming recently. if you have time, take a look at Aditi -- now if i could do that in postgresql i'd be happy.
<7> http://www.cs.mu.oz.au/research/aditi/intro.html
<1> they haven't got any examples of what programming with it are like
<1> do you just submit data and it sorts the rest out
<1> with no idea of a schema or organisation in the user program?
<7> no, it's more like a real relational database system but you can specify constraints using logical clauses like those you would use in prolog. in other words, much more complex rules are possible.
<7> it's a step towards building a generic knowledge maintenance system.
<1> ahh ok
<1> heh are you sure you can't get a postgresql plugin for that? ;)
<1> there a plugins for pretty much everything else under the sun
<7> yes i'm pretty sure. that's why i'm trying to write one.
<1> for fun?
<7> fun is all i do things for.
<7> my website is at http://asmith.id.au fwiw and one of my larger projects is at http://tracktype.org
<1> http://asmith.id.au/ally-testing.html heh your desk looks far more organised than mine ;)
<1> a stack of tidy papers!!!
<7> yeah, i'm a natural born organiser. must be why i'm so drawn to database systems.
<1> im the opposite, but even though my room looks like a tip their are subtle rules at work
<7> you mean like wearing ties and personal hygiene?
<1> like finding my current book will involve working out what the last task I did that made me have to put it down, so it will be under a pile of dirty clothes or in the fridge :)
<7> lol. you must have a very creative mind then. ;)
<7> i still have to do that even when it is the only thing on the desk in front of me. spend too much time with my mind turned inwards.
Return to
#ai
or
Go to some related
logs:
dt_textrel gentoo
ubuntu panel already open
#math
ignasoft media player
netmonk
ubuntu cinelerra freezes
akiga voip
acx unavailable scan
ubuntu could not open the device file /dev/nvidiactl
Python strobject
|
|