| |
| |
| |
|
Comments:
<0> hey everyone
<1> hi
<2> hi
<3> hi
<4> Yes
<4> So has anyone seen umm_lemme/yaroslavB?
<1> not lately. probably on vacation
<4> DanF_DrC Are you one of his acolytes?
<1> hehe no more like his antithesis
<1> I shine light where he doesn't want to see and he reacts like an animal in a trap with curious absence of higher brain functions despite his otherwise good intellect
<4> wow...well put
<4> It seems to me that
<4> if anything, bayesian learning takes place only shortly within working memory
<4> its the very "front end" of the sensory pipe
<1> what learning..
<4> well, yaroslav is a fanatic about bayesian learning
<4> he doesnt seem to know much outside of it either
<4> I can't knock it
<1> what invalidates bayes so far at least is; limited to discretely valued input, the number of inputs can only be a very small number (say <<30) and it has no practical learning algorithm for anything non trivial. Either of these 3 are killer failures
<4> its essentially how some of the DARPA cars work
<4> i dont think its completely invalid
<1> trust me
<4> Certainly it is not a complete depiction of intelligence. Not by far. But its part of part of the whole picture in some way.
<1> so far I see nothing of value in it
<1> any function of the darpa cars handled by a bayesian net is without doubt a trivial function better handled in another way
<1> ***uming any of them even use it
<4> well one thing i think bayes misses is the way in which real animals use elimination.
<4> There is not only a location of an object...but there is the "not location" of an object
<4> wait..what do you mean by a bayesian net?
<5> I like make computer programs that generate music
<5> ...to make...
<4> heh\
<4> what is your method?
<5> i made one program that makes a random walk inside a frame of musical rules, with some repitition built in, became quite ok
<5> i have some samples here --> http://www.randomusic.com/samples/
<4> ok wait
<4> well
<4> that sounds like a cat walking on a keyboard
<4> let me give you some ideas
<5> not really, especially not version 1.11
<5> listen to 1.11 first
<4> ok ok
<4> yes that almost sounds like something
<4> i can really tell that it is still random though
<4> consider giving the algorithm a pre-made theme
<4> and then have it play around with that
<4> are you familiar with mozart's music at all?
<5> is mozart a computer?
<4> no i mean the austrian from the 1700s
<5> is austria a city?
<4> how about bach?
<5> yea how about him, do you like his music?
<4> sure
<5> good
<4> there are really mechanical ways to write baroque music
<4> for example
<4> have the computer generate a "theme" which is just a collection of notes about the same tempo
<5> mozart made a "computer program" that produces random music, it is very nice
<4> then have it play the theme on top of itself , but shifted by some number of notes
<4> hmmm..
<4> *that* didnt go well
<4> Anyone here?
<1> yes
<4> ok danf
<4> to begin where i was before
<4> have the computer generate a "theme" which is just a collection of notes about the same tempo
<4> then have it play the theme on top of itself , but shifted by some number of notes
<4> of course that is one octave lower
<4> have the computer compare all given shiftings
<4> for each shifting, count the number of "harmonious" couplings
<4> where a third or 5th or 6th is created
<4> after all this searching have the program output the best theme
<4> it will have the most harmonious couplings
<1> not sure how that would work. if you can code you can put a test together. a tone is just a sine in its simplest form
<4> DanF_DrC if you study baroque music its mostly created very mechanically
<4> In fact
<4> mozart said about bach's music: "Its hopelessly contrived."
<4> which is probably what jazz musicians say about any music that is not written on the spot
Return to
#ai
or
Go to some related
logs:
yum No Match for argument: spamassassin
#web
jarnhus
opinion, qingy
grub alternative x11.conf
Ubuntu ncurces
ACCESS VIOLATION SUMMARY emerge webapp-config
#python
+perl +Encode +utf8 +install
y2pmbuild
|
|