| |
| |
| |
|
Page: 1 2 3
Comments:
<0> hi
<1> hello!
<0> any news?
<2> <Licky@ef> hey any of you guys try PAULA yet?
<0> Licky: url?
<2> <Licky@ef> http://thepiratebay.org/tor/3391980/P.A.U.L.A._-_the_artificial_intelligence
<2> <Licky@ef> http://www.paoloentertainment.com/software/main.htm
<2> <Licky@ef> oppps
<3> is that the laptop talking mouth chatbot?
<2> <Licky@ef> supposed to be a learning bot
<2> <sheer@ef> is she pretty good?
<2> <Licky@ef> i only have 17% of it so far ;P
<3> looks like malware
<2> <Licky@ef> paula stand for the guy who made it
<2> <Licky@ef> not a chick
<3> should be U.S.E.L.E.S.S. instead
<2> <Licky@ef> P.A.U.L.A stands for Paolo's Artificial Intelligence Limited to ***imilation
<0> hehe
<2> <Licky@ef> i spent all day looking for somthing im happy i found something
<2> <Licky@ef> i can send you a copy when i get all of it we can learn together thats it u.s.e.l.e.s.s.
<2> <Licky@ef> id like to check out ulta hal its supposed to learn also
<0> now that one is for sure useless
<2> <Licky@ef> P.A.U.L.A features several tools to make your virtual human being just the way you want it to be
<2> <Licky@ef> big tits , tight ***
<2> <Licky@ef> An Artificial Intelligence engine, simulating a human being. It learns, understands and feels. Emotions like happiness, anger, and even ***uality are just some of the simulations P.A.U.L.A can perform. You'll be amazed she's not real!
<2> <Licky@ef> i bet we wont be amazed LOL
<2> <Licky@ef> it learns thats all im interested in
<2> <Licky@ef> its only like 30 megs , im already starting to doubt ;P
<2> <Licky@ef> now you all say "Thanks great licky for bringing light to our dark dark AI world"
<2> <Licky@ef> 30%
<4> heh, Licky, you're still on this silly "chat bot" thing?
<2> <Licky@ef> yeah im a taurus im stubborn
<2> <Licky@ef> seriously i have a great irc chat bot once wish i save the code
<5> AIrelay <Licky@ef> big tits , tight *** hahahah
<2> <Licky@ef> ;P
<5> yeah that is the ultimate in AI gynowhorebots :<
<5> well where the funding will come from anyway :[
<2> <Licky@ef> well i found the full version on a site and im gonna try it out ..might be kewl
<5> like how the good parts of the internet are subsidised by pron
<2> <Licky@ef> honestly i heard about 2%
<2> <Licky@ef> 40% meaningless blogs ;P
<5> lol yeah I guess it is now
<4> anyone here have some experience with neural networks for regression?
<4> (specifically I'm working with feed-forward networks with linear outs and a single hidden layer)
<2> <Licky@ef> they said there are 200 million blogs not updated anymore LOL
<1> hi toad, what is your question?
<5> yeah I have a live journal from when it started out that hasn't been edited more than twice
<1> (i know little about this type of linear output network, but maybe i will get lucky and be able to help)
<4> zizazze: I'm just curious about any useful techniques for estimating how many nodes can go in the hidden layer before it starts to overfit, given a training set of N points
<4> (other than just doing cross-validation and trying a bunch of values)
<1> sorry, i only have experience with what you put in (...)
<5> you could use a GA to find out for you :)
<4> Ulti: I doubt a GA is any faster than CV :)
<1> toad, have you ever heard of widrow's rule of thumb?
<4> nope, I'll google it though
<1> the rule of thumb may get you in the ballpark
<1> i use it for a starting value
<1> then i use the procedure you put in (...) to try values larger and smaller than that starting value (mostly larger, in my case, but it may be different for you)
<4> cool, thanks
<1> the version of it i remember hearing is to have one parameter (weight) for every 10 training points
<1> i am not sure if that is the standard form of the rule of thumb, though
<4> yea I've heard anywhere between 10, 30-40, and 100 :)
<1> wow
<1> let me check how many i am using right now
<1> i am actually using 1.25, but i am using early stopping during network training
<1> without the early stopping i wonder if that would backfire
<1> i think probably yes
<1> out of curiosity, what is your data?
<4> derivative data from the Netfilx prize
<1> "Who Wants to Be a Millionaire" ;-)
<4> Only harder to win :)
<1> true
<6> hello, I'm looking for some sort of algorithm similar to Rete, but sends both activation and deactivation messages to the final productions.
<6> instead of just "fire event"
<7> Pennywise: are you there?
<8> you're suppose to say "penny for your thoughts"
<5> http://www.carrotkid.com/android207video.html
<3> rather good
<5> :>
<3> did you make it?
<2> <Licky@ef> hey dudes this paula program at least learns
<2> <Licky@ef> not to bad
<3> still useless I would imagine
<2> <Licky@ef> anyone want to try it out
<2> <Licky@ef> i can send it to ya
<3> no thank you
<9> comments&criticism on this article would be very much welcome, ainerd.com/elfrte.pdf
<9> (the source code is available at ainerd.com/opthol.tar and ainerd.com/probqa.tar but it's 300 MB)
<10> 300M of source!
<9> hehe okay not just source
<9> it also contains various statistical language models, that make up the main body of the space
<9> (MXPOST, MaltParser, MSTParser and the stanford log-linear pos tagger are included)
<9> (and treetagger)
<0> hey ho
<11> hi, all
<11> I need a small AI system for implementing CaptureTheFlag in 2D on MSVC like OpenSteer. Does anybody know any free like this?
<0> doesn't opensteer work?
<11> AJC i tried to imlement it in Lightfeather 3D engine, but after just adding files to project MSVC throws 56 errors on link stage...
<0> what kind?
<0> doesn't sound like it's opensteer's fault, does that build as provided?
<0> (this is just typical of C++ btw :-)
<12> you probably need to use a different compiler
<12> or different version of MSVC
<0> i'm pretty sure that MSVC 2003 (7.1) or 2005 (8.0) would be supported out of the box
<0> bethrezen: ping
<2> <Licky@ef> hey nut squads
<2> <Licky@ef> that paula program i got was crap LOL
<2> <Licky@ef> glad i didnt pay 30 bucks for it LOL
<3> told you : )
<0> Licky: what did it do?
<2> <Licky@ef> i can send it to ya if ya want to check it out
<2> <Licky@ef> it learns a bit but i didnt mess with it long
<0> on way, just describe it roughly
<0> no way i mean
<3> a chatbot
<3> is my guess
<0> "you like to chatbot?"
<2> <Licky@ef> i had the web site
<2> <Licky@ef> it learns and talks and has a male or fdemale avatar
<2> <Licky@ef> i programed an awsome chatbot using mirc figured out how to program irc found all the right variables to use and it was nice
<2> <Licky@ef> then i lost the code when my harddrive died
<2> <Licky@ef> thinking about starting a new one
<2> <Licky@ef> you know any good ones out now i just got mirc 6.21 cracked
<2> <Licky@ef> i had it so it could go on any channel and listen/learn and not say anything
<2> <Licky@ef> it saved everythiong it learned
<2> <Licky@ef> it was HOT ;P
<2> <Licky@ef> it could have easily been as smart as an ant ;P
<2> <Licky@ef> AJ what are some good chatbots?
<2> <Licky@ef> ill use one to start from
<2> <Licky@ef> there are some mirc variables you use to save to a file and if i can remeber ill be on my way
<2> <Licky@ef> brb
<13> anyone knows a good & free speech recognition framework?
<1> sure
<1> check out the Freeware list here: http://isca-students.org/Resources
<1> Sphinx and HTK are popular. Check the license terms to make sure they suit you. I think the Sphinx license is nonrestrictive, but I am not sure. If you don't like either license there are other frameworks available.
<13> cool ty
<14> licky: mirc? cracked? there is a specification of the irc protocol and interfaces for several languages, such as java, perl and ruby
<2> <Licky@ef> really
<2> <Licky@ef> so if i have 6.21 cracked i can use perl or whatever with it?
<14> IRC and mIRC are two different things, one is a protocol and the other an user interface
<2> <Licky@ef> well its not cracked it just has a keygen ;P
<2> <Licky@ef> and it is 6.21
<14> if you want a bot to be on IRC, you just use the IRC protocol
<14> mIRC is just a program for windows that use IRC
<14> you do not *need* mIRC to use IRC, i use xchat for instance
<14> the reason you may want to "crack" mIRC would be to spy on other people using mIRC on your computer
<14> i.e. if they sit in front of your computer, not on other computers
<14> if you just want a bot to follow this conversation, you throw up a TCP socket and log on using the IRC protocol
<2> <Licky@ef> do you use a chatbot?
<2> <Licky@ef> what you said earlier is, if you use the right interface you can also use pearl ,java in irc?
<2> <Licky@ef> what interface
Return to
#ai
or
Go to some related
logs:
dguitar startup ubuntu
#lgp
#gentoo
sudo dhcpclient
#lgp
migrate from gentoo ubuntu
gentoo broadcom gigabit dhcpcd
#perl
#perl
#kde
|
|