| |
 |
|
|
Science Forum Index » Compression Forum » Help with WFC Implementation
Page 1 of 1
|
| Author |
Message |
| Arsène Lupin |
 Posted: Wed Feb 27, 2008 4:21 pm |
|
|
|
Guest
|
Hi, people!
I wanted to compare algorithms for the second stage, and for that I'm
using the algorithms TimeStamp, MTF, MTF-1, MTF-2 and WFC. I managed
to create an algorithm that performs better than the ones I'm testing,
even WFC. The problem is, comparing with the Deorowicz implementation
from his paper, the compression comparition of my algorithm is far
worse, because my last step is a single order-0 encoder (the fpaq0p
from Matt Mahoney). Moreover, I could not do the speed improvement he
teaches, and my function is based on a work of 2001, not on the 2003
one (from his dissertation - I could not understand it ~_~).
If any of you had a source code, a program to generate only the second
step transformation, or even the Calgary Corpus files compressed with
the BWT and the WFC to send me, I'd be glad. I tried to reach
Deorowicz by mail, but the only references to his mail are outdated.
As soon as I'm able to compare them right (using the same parameters),
and I manage to prove (or not!) that my algorithm performs better,
I'll post it here, so however who wants can do improvements (that I'm
certain are possible). I'll publish it for my dissertation, so a paper
should be avaliable soon.
If any of you want the code of WFC I already have, I can it post too,
so we can work with something already created.
Thanks for all the help.
Arsène |
|
|
| Back to top |
|
| biject |
| Posted: Fri Feb 29, 2008 8:08 am |
|
|
|
Guest
|
On Feb 27, 7:21 pm, "Arsène Lupin" <deten...@gmail.com> wrote:
Quote: Hi, people!
I wanted to compare algorithms for the second stage, and for that I'm
using the algorithms TimeStamp, MTF, MTF-1, MTF-2 and WFC. I managed
to create an algorithm that performs better than the ones I'm testing,
even WFC. The problem is, comparing with the Deorowicz implementation
from his paper, the compression comparition of my algorithm is far
worse, because my last step is a single order-0 encoder (the fpaq0p
from Matt Mahoney). Moreover, I could not do the speed improvement he
teaches, and my function is based on a work of 2001, not on the 2003
one (from his dissertation - I could not understand it ~_~).
If any of you had a source code, a program to generate only the second
step transformation, or even the Calgary Corpus files compressed with
the BWT and the WFC to send me, I'd be glad. I tried to reach
Deorowicz by mail, but the only references to his mail are outdated.
As soon as I'm able to compare them right (using the same parameters),
and I manage to prove (or not!) that my algorithm performs better,
I'll post it here, so however who wants can do improvements (that I'm
certain are possible). I'll publish it for my dissertation, so a paper
should be avaliable soon.
If any of you want the code of WFC I already have, I can it post too,
so we can work with something already created.
Thanks for all the help.
Arsène
I know your seem to be testing various methods. But since your
asking
for code or full model of WFC could you fully describe your method
or
provide code for the method your using.that seems to beat everything
your testing. And what do you get using yours for the calgary corpus.
David A. Scott
--
My Crypto code
http://bijective.dogma.net/crypto/scott19u.zip
http://www.jim.com/jamesd/Kong/scott19u.zip old version
My Compression code http://bijective.dogma.net/
**TO EMAIL ME drop the roman "five" **
Disclaimer:I am in no way responsible for any of the statements
made in the above text. For all I know I might be drugged.
As a famous person once said "any cryptograhic
system is only as strong as its weakest link" |
|
|
| Back to top |
|
| Arsène Lupin |
| Posted: Sun Mar 02, 2008 2:04 am |
|
|
|
Guest
|
Thanks, David.
But the reason why I asked for help to construct the WFC algorithm
(the better implementation, because I already have one) is that I
don't want to "brag" about something that I still didn't achieved. For
some reason this group doesn't like to exchange source code (there is
a few exceptions, though).
I wanted a fast and reliable way to go straight to the testing, but
since I got nothing, what I have to do is what I have to: Find a way
to implement it myself.
Thanks a lot.
Arsène |
|
|
| Back to top |
|
| |
|
Page 1 of 1
All times are GMT - 5 Hours
The time now is Fri Oct 10, 2008 7:36 pm
|
|