On Wednesday 28 February 2007 9:27 am, Andrew Burgess wrote:
> >I was referring to the definition of a "heuristic". Quoting Wikipedia,
> >the relevant part:
>
> > Heuristics are intended to gain computational performance or
> > conceptual simplicity, potentially at the cost of accuracy
> > or precision.
>
> >(Or more classically, trading off correctness in some cases to
> >achive simplicity ... mostly when a correct algorithm would be
> >a real PITA to implement; or is too slow; or hasn't been defined.)
>
> I always thought it refered to a method obtained through
> trial and error, nothing to do with (or orthogonal to)
> inaccuracy.
I didn't like the Wikipedia discussion all that much, which is
why my addendum talked about correctness not accuracy. I came
across academic definitions in two contexts:
- Artificial intelligence, which is aligned with your
comments ... "correct" hasn't been defined for all cases,
and in some cases it might not be definable. For example,
"expert systems" sometimes include large collections of
heuristics that must be balanced against each other.
- Algorithm analysis, where there may be algorithms which
are always correct, but may be e.g. O(N!) rather than a
simpler heuristic that's O(N) but which still fails less
often than the chip used to run the algorithm.
By your comments, I think you're more aligned to the former
(AI) perspective. But I always thought the latter was an
interesting point, and came across it again recently in the
context of numerical error analysis (FP stuff) ... where it
can be better to tolerate more errors in some contexts, when
the alternative is that important boundary cases would become
unusably broken. (And where intermediate results may sometimes
seem surprisingly incorrect... the macro result being important,
the micro steps less so.)
- Dave
p.s. Returning briefly to the topic, defining a parameter
in a specification can be arbitrary, and its specific
value may have been chosen heuristically, but once
chosen that becomes the definition of "correct". :)
> Several dictionaries do not refer to inaccuracy:
> http://dictionary.reference.com/browse/heuristic
>
> Just OT and FYI
>
> Cheers
> Andrew
>
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
linux-usb-devel@lists.sourceforge.net
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
