8/13    Is there such a thing as an algorithm that can output another
        algorithm? Kind of like self tuning, self evolving algorithm?
        \_ Yes. Not as glamorous as it sounds; see also: genetic algorithms,
           Remez algorithms, data-directed programming
        \_ bison, yacc, (f)lex, and many many more. not self tuning though.
        \_ There are algorithmns that can learn from data.  It's really not
           as sexy as some people seem to think, just an application of
           statistics.
        \_ self-propagating neural networks
           \_ Neural nets are data structures not algorithms.  And
              there is no such thing as a self-propagating neural net.
              There are neural nets with feedback loops, but they don't
              output 'another algorithm.'  They have some uses in time
                        \_ Tell us of the stars....
              series prediction.  Please don't be an idiot, read Russell's
              book or something.
              \_"A neural network is a processing device, either an algorithm
              or actual hardware, whose design was motivated by the design and
              functioning of human brains and components thereof" from the NN
              FAQ. And I believe it is theoretically possible for a NN to
              output another algorithm. Sign your posts. williamc
                \_ A NN FAQ is not an AI authority.  For one, this "FAQ" seems
                   to imply that an algorithm is a processing device, which
                   is idiotic.  And a neural network cannot 'theoretically'
                   output an algorithm.  It outputs an array or list of
                   floating point numbers.  I recommend Christopher Bishop's
                   neural networks book as an excellent way of learning more.
                \_ It's easier to babble on the motd and toss buzzwords
                   around.
                \_ how dare you call me a "data structure". -brain
        \_ feedback loops
        \_ Does self-modifying code count?
        \_ Optimising compilers fit, too...
        \_ and my thesis!