home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.ai.neural-nets:3216 comp.ai:3131 comp.theory:1774
- Newsgroups: comp.ai.neural-nets,comp.ai,comp.theory
- Path: sparky!uunet!wupost!gumby!destroyer!ubc-cs!fornax!jamie
- From: jamie@cs.sfu.ca (Jamie Andrews)
- Subject: Re: measuring complexity (was Re: Kolmogorov Complexity)
- Message-ID: <1992Aug17.171934.9364@cs.sfu.ca>
- Organization: CSS, Simon Fraser University, Burnaby, B.C., Canada
- References: <1992Aug12.150857.219@access.digex.com> <BsxH2n.Ks4@math.uwaterloo.ca> <1992Aug14.183256.20441@access.digex.com>
- Date: Mon, 17 Aug 1992 17:19:34 GMT
- Lines: 32
-
- As far as I can tell, there is no mathematical reason why
- K-complexity for learning applications should be based on any
- one computational formalism (Turing machines, assembler code, C,
- etc). However, for the sake of consistency of notation, it
- might be better to base it on some programming language with a
- simple and standard semantics, such as pure LISP or pure Prolog.
-
- So, for instance, the K-complexity of a given list of
- numbers might be defined as: the length of the shortest program
- in pure LISP such that the value of the s-expression "(result)"
- is that list. (Well, I guess it would have to be pure LISP plus
- some arithmetic functions.) The "length" of a program might be
- defined as the total number of atoms plus the total number of
- sets of parentheses, to give a more realistic idea of the
- storage space needed by the program.
-
- A measure of K-complexity relative to some standard suite
- of function definitions could be given, too. The relative
- K-complexity would be the length of the shortest pure LISP
- program containing that suite of definitions, _minus_ the length
- of those definitions. The minimal program could then make use
- of the "knowledge" in the standard functions without having to
- derive it.
-
- By the way, I asked about K-complexity here a few years
- ago, and was given convincing arguments that finding the
- K-complexity of a given string is in general undecidable.
- I don't know about randomized algorithms, though.
-
- --Jamie.
- jamie@cs.sfu.ca
- "Every \item command in item_list must have an optional argument." LaTeX pg.168
-