Path: utzoo!utgpu!watmath!att!pacbell!ames!sun-barr!cs.utexas.edu!csd4.milw.wisc.edu!mailrus!cornell!vax5!b27y From: b27y@vax5.CIT.CORNELL.EDU Newsgroups: comp.ai Subject: Re: evolution of intelligence Message-ID: <19266@vax5.CIT.CORNELL.EDU> Date: 7 Aug 89 18:51:34 GMT References: <2153@hub.UUCP> <19229@vax5.CIT.CORNELL.EDU> <591@berlioz.nsc.com> <5768@pt.cs.cmu.edu> Sender: news@vax5.CIT.CORNELL.EDU Reply-To: b27y@vax5.cit.cornell.edu (Michael Gray) Organization: Cornell Information Technologies, Ithaca NY Lines: 37 In article <5768@pt.cs.cmu.edu> jps@cat.cmu.edu (James Salsman) writes: >..... Exercise: what is the optimal structure of an >AI researcher's neural-net. > >Also, by "size of brain" do you mean > (A) Weight > (B) Volume > (C) Number of Neurons > (D) Number of Connections > (E) Resolution of each Connection > (F) Rate of Connection Change > (G) Number of desernable Neurotransmitters >or some combination of these, or somthing else? > >:James > While i have only dabled in Neural-nets. i am nor sure if there is any real measure yet for neural-nets for this: specialization of functional units: Minksy loves to talk about these thing (_Society of Mind_), and modern brain physiology seems to indicate that intellegent systems have several functional compenents - each with its own tasks, yet all very highyl connected to together. I seem to recall that it is the reason that we are supposed to be smarter than whales(?), yet have smaller brains (I admit that the last sentence is a subjective statement, and maybe wrong). I am sure that some NN guys might like to think that each "neuron" is a functional unit, but I feel the neccesity to restrict my self to abstract functions like :speech, vision, etc.. /*----------------------------------------------------------------------*/ /* Michael Gray b27y@vax5.cit.cornell.edu (ARPA) */ /* Misha Computing b27y@crnlvax5 (BITNET) */ /* 526 Stewart Ave UUNET!CORNELL!VAX5!b27y (UUNET) */ /* Ithaca N.Y. 14850 (607)277 2774 */ /* */ /* "Save the Humans" --Bumper Sticker */ /*----------------------------------------------------------------------*/