Loading ...
Sorry, an error occurred while loading the content.

Trivial stuff about learning

Expand Messages
  • David_Augusto_Villa
    wizard_of_frozzbozz wrote: > one has to consider more factors then just the > volume of material the software is able to > learn . One
    Message 1 of 1398 , Oct 28, 1999
    • 0 Attachment
      wizard_of_frozzbozz wrote:<br><br>> one has to
      consider more factors then just the<br>> volume of
      material the software is able to<br>> "learn". One must
      also consider can the software<br>> "use" this
      "knowledge" in a meaningful way<br><br>I think this is an
      excellent response. Learning isn't just remembering
      everything you've run across; it's recognizing patterns and
      applying those patterns to other problems.<br><br>Let me
      give a stupid example. We could try to have a system
      learn elementary algebra by storing all of the problems
      and answers it sees. So if you ask it again what x is
      if 5-x=3 it will tell you 2, but it wouldn't
      necessarily be able to figure out what x is if x-3=2. And
      even if the answer is somewhere in the database, it
      might take some time to find it.<br><br>Of course this
      is an incredibly naive approach. But it's difficult
      not to place a bound on what a program can learn.
      Just because a program can solve for x in linear
      equations doesn't mean it can solve problems with absolute
      values, or with quadratics, or whatever the next step may
      be.<br><br>Mathematica may do a better job at solving equations, but its
      extensive collection of techniques was not aquired through
      learning. A program which can figure out how to solve even
      simple algebraic equations would be much more
      interesting, since the technique could be applied to more
      complex problems.<br><br>But this is an open question.
      Even with state of the art AI, to get a program to
      develop algroithms you have to hand-feed it: you have to
      structure the enviornment so that it will inevitably run
      across a solution. That is to say, no one really knows
      how to make a program intelligent.<br><br>> Also,
      it should be able in some way to link<br>> topics
      that are seemingly unrealated.<br><br>It's difficult
      enough to link topics that are seemingly trivially
      related. Addition of integers and addition of rational
      numbers technically overloads the addition operator. In
      fact, in a strict sense, integers are not rational: we
      establish a surjective function from the integers to the
      rationals by dividing by one. But this function is so
      intuitive that no mathematician would argue the fact that
      42 is rational.<br><br>Of course there are more
      complex examples. What is the factorial of 3.5, if that
      even makes sense? What is the measure of a dense
      subset of the unit interval which has no interior
      points? The solutions to these sorts of problems aren't
      intuitive enough to overload operators in practice, but
      there are clearly still conceptual
      connections.<br><br>-David Villa
    • smartxpark
      Hello, Have you succeeded ... God To Create Silicon
      Message 1398 of 1398 , Aug 19, 2005
      • 0 Attachment

        Have you succeeded

        --- In artificialintelligencegroup@yahoogroups.com, Spider_Plant9 wrote:
        > The rosetta stone is a archeoliguist's dream come
        > true. The Stone was found in a place called rosetta
        > from what I remember but it helped solve the
        > hieroglyphics puzzle because on the stone was written the same
        > text in three languages Greek, Egytian, and Latin. We
        > knew both Greek and Latin. The egytian language though
        > unknown was also the same texts as the known languages.
        > So, using the other two languages we were able to
        > piece together egyptian.<br><br>Now, you can guess what
        > I'm going to do with the texts and languages I
        > collect. I'm going to create a computerized version of the
        > rosetta complete with the "thought" processes to use it
        > directly and extrapolate more words and
        > sentenses.<br><br>Spider_plant9<br>Archeology and the "Words of
        God"<br>To Create Silicon
        > Life<br>Anybody know of a set of standardized interpretations for
        > the Qur'an.
      Your message has been successfully submitted and would be delivered to recipients shortly.