| |
| |
| |
|
Page: 1 2 3 4 5 6
Comments:
<0> hi all
<0> can i ask a simple question about backtracking
<1> jc-denton: explain what you mean by backtracking
<0> how to use backtracking for solving sudoku
<2> hey hold on
<2> http://www.norvig.com/sudoku.html
<0> thx
<0> lol
<0> this person did it in python
<0> i wanted to do that too
<0> but i have to learn java for school, so..
<2> well just translate
<2> your school is evil by the way
<0> lol
<0> why?
<2> since you are required to use java
<0> well java is proprietary
<0> heh
<0> well i would be happy if they would focus more on algorithm then java and software engineering
<0> but java is fine as implementation language
<0> i think
<2> im not sure
<2> i not java obscures the algorithms
<0> ?
<2> and is so unwieldly that you need engineering disciple from the get-go
<2> to get anything done
<0> ja i prefer python too
<0> but java works fine (but slow heh)
<3> I dislike java.
<3> I prefer C++.
<3> I feel Java forces the OO paradigm onto things that do not fit it well, and damages the OO paradigm for those that do.
<0> humm
<0> c++ is more difficult i think
<2> c++ > java
<0> but if your skilled at c++, that's good
<2> c++ gives you the option to opt out of the OO paradigm
<2> and templates ist ***y
<3> right.
<3> the strength of C++ is that it allows you to both select the paradigm that fits the problem and mix paradigms.
<0> in python you can code w/o OO too
<0> but of course python uses a vm as java
<0> which makes it slow
<2> but its lack of support for functional features, makes me use things like python over c++
<3> so that if I have a problem where part of it is very easily solved with OO and part of it isn't... (a GUI frontend for some sort of computational backend, say...)
<3> I use C++ over python because I know C++ and not python. I use C++ over java because I hate java >.>
<3> sometimes I use PHP over C++, because there's some tasks I do where PHP is just a better fit, and I'm pretty comfortable with it.
<3> (occasionally I use bash scripting, too. :P)
<2> i find c++ is just too close to the bare metal for me
<0> i prefer python over bash scripting
<0> :D
<0> well i just hate people who think you need a particular language to get anything done
<2> I think a multilingual approach is key
<0> php is quite easy
<2> I love php, everyone's code work so well since everyone uses the same code
<1> i like c++
<1> lo Captain_Fourier
<2> hi
<3> php is awesome for web-based stuff.
<3> c++ is nice if you're comfortable with it, becuase it's so general-purpose it can do whatever you want without much hacking around the language, and there's enough specialpurpose libraries that if you stick with a specific kind of task for a while, you'll be able to shortcut a lot of code.
<1> i did some thinking last night about how to enumerate the number of possible non-looping programs that could match a set of inputs to outputs
<3> oh dear
<1> it seems as though doing it by complexity would make sense
<2> yum
<2> _death, have you heard of marcus hutter?
<1> no, who is that?
<3> http://norvig.com/21-days.html <-- this is amusing.
<3> a lot of that stuff is so true, too.
<2> norvig is da ****!
<1> marcus hutter seems to have done some work in the area
<2> http://www.idsia.ch/~marcus/index.htm
<4> shalom
<1> i must wander off now
<1> later ;)
<5> java would be fine if it could compile to native code and if they would stop making the ****ing IDE in java!
<3> haha
<6> those IDE's are so nice, make them see vim's console and they just can't code.
<2> huh
<2> hey whats a good for less than $30/?25
<7> woot I just redid my NN lib, mabe is about 15 times faster :)
<7> let me make sure it still works
<7> correctly
<7> lol it helps to have a learning rate larger then zero lol
<2> you mean having the ability to learn?
<5> Shadow_mil, sure
<7> lol
<5> Captain_Fourier, no. ANN
<7> I was like, why the hell is it not working!>!
<7> going crazy checking equations
<5> :)
<7> and the I saw $lr = 0;, like 10 mins later
<7> like darn it
<7> then*
<5> I know :)
<7> I think its kinda wrong it call it a reward...
<7> *NN gets it right
<7> great job... NOW REMEMBER IT!!!!
<7> for(int i =0; i<= 1000; i++) train() //MUHAHAHAHAHAHAHAHA!
<7> *NN cries out
<7> how do you determine how many hidden nodes you will need?
<2> hey DanF_DrC whats a good ai book for less than $30/?25
<5> Shadow_mil, guestimation
<7> ah. That soild logic again hun
<7> DanF_DrC: how do you come up with interfaces for robotic agents?
<5> Captain_Fourier, you can try a used one. some like 'AI a modern approach', probably because it's popular in style but more serious in my opinion is Mitchell's 'Machine Learning'
<5> Captain_Fourier, but I would say (and many will disagree and be wrong) that AI is only the parts that relate to neural nets and the 3 paradigms supervised, reinforcement and unsupervised learning. both book contain a lot of lesser techniques as well
<8> what about probability nets?
<5> there might exist a book that hits AI just on the head but I'm not aware of one. no doubt many with tempting titles. the two books I mentioned are the two most used for ML course at unis
<5> SDr, irrelevant and should be avoided
<8> mmm, I haven't seen DanF_DrC pointing out how "it" should be done, rather than *what* should be done, which is, imho, irrevelant in helping others.
<8> it's, like, "hey, how to configure apache?" "you'd rather go play doom4"
<9> heh
<7> ...
<5> the blunt style is not always appreciated :)
<9> i think that you have people who get lost in the world of semantic nets, bayes statistics, and "knowledge representations" and after studying this for years, they still dont have an intuitive idea of how an insect's brain works.
<9> this is lopsided to me.
<9> Then what I usually get is the regular "But AI is not about copying nature!!!"
<9> but that's like saying "I'm not going to use my textbook on this test because it already contains methods!!!"
<5> even in the machine analogy insects are too hardwired for interest
<8> and I think, that we're pursuing different goals in the field of AI -becouse even though they might not have any ideal, at all, about how an insect's brain works, they can solve problems in their own domains
<5> and certainly many will spend time on avenues that will never pan out. like CYC
<8> oh, excuse me for calling cl***ifications, logical tools, and natural language processing AI
<8> how dare I.
<5> NLP in general is likely futile and an entry point. much like a baby doesn't start with language but learns
<7> The best way to do anything type of learning adaptation that we know of is nature
<5> the term AI is defacto ambiguous. for moderate clarification I often use the term real AI
<9> DanF_DrC I always ***umed that CYC people and other "common-sense knowledge" people were still operating under the ***umption that AI is about creating a convincing illusion.
<8> DanF_DrC, yes it does. 80% of the language is wired in the brain -> Stephen Pinker: Language instinct.
<7> ANNs and GAs are both rooted in natural ideas
<5> paros, quite but unconsciously. they won't admit it to themselves
<9> DanF_DrC There is a series and widespread PROBLEM here you know.
<5> pinker has funny hair and looks like he could be from another planet. other than that he has little appeall
<9> Namely, that when people read what Alan Turing wrote about his
<5> -l
<9> 'Turing Test' they think that means creating an *illusion* that can trick researchers.
<9> This problem should be brought out into the open and clarified more often.
<5> yep the turing test is misleading for the newcomer. but if one can't navigate past that one is not likely to succeed anyway
<7> paros: Well the thing that I have a problem with is, even if you create a AI that can learn, it can only learn with in its policy you have given it.
<8> DanF_DrC, would you please be so kind, to point out the mistakes he publicated in his books?
<9> Someone in THIS CHANNEL once linked me to a chat bot a la loebner prize.
<9> And told me the person said they had "created the first AI"
<5> SDr, I wouldn't read it. you can mention what you feel is his most valuable contribution and I will gladly consider it
<8> Becouse it looked to me something, that has way more verifiable, repeatable, and falsificatable predictions, than I have ever heard from you.
<9> I have to tell you , Dan, this makes me put my face in my hands...shake it back and forth and have a long...exhausted sigh.
<5> paros, don't dwell on it
<9> DanF are you familiar with Gerald Edelman of Rolf Pfeifer?
<5> no
<9> or*
<7> paros: I do not think we should give it a turing test, but a human style IQ test.
<8> okay then; I think I'm going back to my futile work of researching useless junk.
<9> DanF_DrC You need to start reading thier books.
Return to
#ai
or
Go to some related
logs:
TIACX100C0d
libcurl was built with HTTP disabled
#web
problem pyhook encoding
atftpd cant bind port 69
install SABnzbd ubuntu
ndiswrapper division by zero
gcursor cant doesnt
fglrx portrait xorg.conf
ubuntu ensoniq mic
|
|