Re: Hutter Prize Rules
- From: "James A. Bowery" <jabowery@xxxxxxxxx>
- Date: 24 Aug 2006 08:50:39 -0700
Ah, I missed this statement of yours:
Michael Goldshteyn wrote:
but I am willing to state unequivacally that no program
will be able to compress the original 18,324,887 bytes by more than an
additional 32% of the just mentioned already compressed size, thus making
that pretty much the asymptote for all future attempts. I base this on the
at least 1-bit per character characterization of the data which has oft been
attributed to English text,
Shannon himself was not so unequivocal when he put the range at between
0.6 and 1.3 bits per character -- but that was based on human
performance. Some people state unequivocally that computers won't
achieve in the human range but if paq8hp1 is validated it is already at
1.42 bits per character.
BTW: The history of text compressors has been to acheive about 3%
improvement per year. If paq8hp1 is validated we will have 3%
improvement just since the announcement of the Hutter Prize.
At this rate, the top end of Shannon's estimate of human performance
will have been reached within months of the announcement of the Hutter
Prize.
.
- References:
- Hutter Prize Rules
- From: Marcus Hutter
- Re: Hutter Prize Rules
- From: Michael Goldshteyn
- Re: Hutter Prize Rules
- From: James A. Bowery
- Re: Hutter Prize Rules
- From: Michael Goldshteyn
- Hutter Prize Rules
- Prev by Date: Re: ADPCM Source Code in C
- Next by Date: Re: Lossless/Near lossless, Bayer/Pixel-shift and Noise removal advice for community oriented Digital Cinema Camera projects?
- Previous by thread: Re: Hutter Prize Rules
- Next by thread: Re: Hutter Prize Rules
- Index(es):
Relevant Pages
|
