The paper "A Bijective String Sorting Transform"
biject
it gets accepted. I don't know so far I have batted zero on any paper
actually being published with my name on it. I actually think the
transform has more application in crypto than compression. But there
are niche places where it will do very well for compression time will
tell.
No matter how far his goes I would like to thank Yossi for trying to
follow my rantings on the phone while he was thousands of miles away.
I would lilke to thank Mark Nelson who planted the dream of this many
years ago. There are many others most of whom I never meet and most
likely never will.
I don't like being in public but if the paper is accepted I guess I
have to buy a tie and a clown suit so
I can mingle with people.
Thomas Richter
Some comments:
o) Please don't say that the BWT is not invertible as people will object. Instead, say that your
transformation is invertible even in the absense of an EOF indicator, which would be more in
the traditional line of thinking.
o) Depending on the conference you want to file this to, you need to consider the page limit.
On the DCC (data compression conference), which I can recommend very much, you do have ten pages,
but for many other conferences, the limit is six or even four pages, please check this, and please
check the page format.
o) You don't need to buy a tie or a clown suit. I hate ties, and had no problems coming without
one. (-;
Otherwise, it's a nice work from my perspective, and I don't see why it shouldn't be accepted, keeping
1) in mind: Don't try to stomp on peoples feet unnecessary - if you get what I mean.
So long,
Thomas
Tom St Denis
> o) Please don't say that the BWT is not invertible as people will object. Instead, say that your
> transformation is invertible even in the absense of an EOF indicator, which would be more in
> the traditional line of thinking.
Agreed was going to point out the same thing. Well more so that you
don't need an EOF marker just the index of the original phrase.
> o) Depending on the conference you want to file this to, you need to consider the page limit.
> On the DCC (data compression conference), which I can recommend very much, you do have ten pages,
> but for many other conferences, the limit is six or even four pages, please check this, and please
> check the page format.
The problem at least so far as sci.crypt is concerned is that the work
has no cryptographic significance and would not get accepted at any
crypto conference, regardless of paper quality. It's better suited to
a data or text compression conference.
> o) You don't need to buy a tie or a clown suit. I hate ties, and had no problems coming without
> one. (-;
>
> Otherwise, it's a nice work from my perspective, and I don't see why it shouldn't be accepted, keeping
> 1) in mind: Don't try to stomp on peoples feet unnecessary - if you get what I mean.
I'd like to see them speak to speed/memory usage by comparison to a
normal straight up BWT sort. I've implemented BWT routines in some
rather tight spots before (like on a Gameboy) and would be interested
to see how well their work does. Ultimately, raw compression ratios
are not the be-all of getting people to accept, let alone use, a new
algorithm. bzip2 for instance is widely deployed. Switching it over
to the S transform may seem like a no brainer to the engineer, but is
a 2% gain really worth causing never ending compatibility problems
with people all over the world?
Specially since LZMA (which is a fairly standard Linux utility
nowadays) beats bzip2 by more than 2%?
And yeah, you don't need to dress up for most academic conferences.
They're pretty laid back from what I've seen. If you get accepted
just go and enjoy yourself, keep some of your more contended opinions
to yourself, etc...
Tom
Thomas Richter
> The problem at least so far as sci.crypt is concerned is that the work
> has no cryptographic significance and would not get accepted at any
> crypto conference, regardless of paper quality. It's better suited to
> a data or text compression conference.
Exactly so. This is why DCC would be an excellent place. Not a cheap
one, though.
>> o) You don't need to buy a tie or a clown suit. I hate ties, and had no problems coming without
>> one. (-;
>>
>> Otherwise, it's a nice work from my perspective, and I don't see why it shouldn't be accepted, keeping
>> 1) in mind: Don't try to stomp on peoples feet unnecessary - if you get what I mean.
>
> I'd like to see them speak to speed/memory usage by comparison to a
> normal straight up BWT sort. I've implemented BWT routines in some
> rather tight spots before (like on a Gameboy) and would be interested
> to see how well their work does. Ultimately, raw compression ratios
> are not the be-all of getting people to accept, let alone use, a new
> algorithm. bzip2 for instance is widely deployed. Switching it over
> to the S transform may seem like a no brainer to the engineer, but is
> a 2% gain really worth causing never ending compatibility problems
> with people all over the world?
Well, this is after all research. If it is well-done, then a conference
will at least provide some feedback from the real experts in that field
(and I'm not a BWT expert), such as whether the algorithm is probably
already known, etc... Thus, even in case it is *not* accepted, you get
good feedback in the reviews. The O.P. should only speak in the same
language as the peer reviewers.
> Specially since LZMA (which is a fairly standard Linux utility
> nowadays) beats bzip2 by more than 2%?
I'd say that this is a different question. Not an irrelevant question,
but the second question to be asked later.
> And yeah, you don't need to dress up for most academic conferences.
> They're pretty laid back from what I've seen. If you get accepted
> just go and enjoy yourself, keep some of your more contended opinions
> to yourself, etc...
Exactly. (-;
So long,
Thomas
Tom St Denis
> Well, this is after all research. If it is well-done, then a conference
> will at least provide some feedback from the real experts in that field
> (and I'm not a BWT expert), such as whether the algorithm is probably
> already known, etc... Thus, even in case it is *not* accepted, you get
> good feedback in the reviews. The O.P. should only speak in the same
> language as the peer reviewers.
There are many ways to modify BWT [or any codec] in seemingly
desirable ways. The only way to determine if it's actually useful is
to show how it improves or is otherwise novel over other designs.
Just being bijective isn't really enough in either the crypto or
compression world to be novel enough to spend time thinking about. If
it's just as slow [or memory intense] as LZMA but compresses worse,
than who really cares?
> > Specially since LZMA (which is a fairly standard Linux utility
> > nowadays) beats bzip2 by more than 2%?
>
> I'd say that this is a different question. Not an irrelevant question,
> but the second question to be asked later.
It's the sort of thing I'd expect to see from people proposing a new
codec for people to examine. It's kinda the whole point of new
research though, either to come up with a radically different idea
that over time may be optimized, or to radically improve upon existing
work.
As far as I can tell the S transform while neat in its own right fails
both tests. It's neither entirely original work, nor so far as I can
discern radically more efficient than existing work. It's merely
different. And it's up to them to make their case for the design
which they haven't really done in the paper. The only selling point
[so to speak] they've made is that it compresses marginally better
than a standard block sort, while entirely side stepping speed and
memory.
If that's all I have to go on I'll just point them to LZMA and dismiss
the invention out of hand. This isn't mean to be mean it's meant to
be pragmatic. They should look at it from the opposing point of view,
that is, why should I care about this research?
Tom
Ilmari Karonen
>
> And yeah, you don't need to dress up for most academic conferences.
In my experience, at least unless you happen to be one of the few most
established and respected experts in the field, you _are_ generally
expected to show up wearing _something_.
*g,d&r*
(Sorry, couldn't resist.)
--
Ilmari Karonen
To reply by e-mail, please replace ".invalid" with ".net" in address.
Unruh
>On 2009-07-10, Tom St Denis <t...@iahu.ca> wrote:
>>
>> And yeah, you don't need to dress up for most academic conferences.
>In my experience, at least unless you happen to be one of the few most
>established and respected experts in the field, you _are_ generally
>expected to show up wearing _something_.
Which was why the "up" was in that sentence, although a hypen might have
been helpful.
Glenn
are accepted as papers. Also DCC's focus -- its technical scope -- is
narrower than most.
Also, bring your skis because the best snow in the world is right
outside the conference room (if you get bored with the wavelet
stuff...)
Thomas Richter
> If you want to present at DCC, keep in mind that only about 44 papers
> are accepted as papers. Also DCC's focus -- its technical scope -- is
> narrower than most.
However, it is a high-rated conference, so if it gets accepted, you have
a good publication, and if not, you got excellent feedback. I would
believe that this work fits pretty well into the focus of the conference.
> Also, bring your skis because the best snow in the world is right
> outside the conference room (if you get bored with the wavelet
> stuff...)
We haven't had much wavelet stuff in the last years, actually.
Multi-channel coding seems to be one of the active areas.
Nevertheless, if you do ski, definitely bring your skis, yes. (-:
So long,
Thomas
mk
to read my paper on the bijective sort transform.
http://www.informatik.uni-stuttgart.de/fmi/ti/mitarbeiter/Kufleitner/publications/kuf09psc.pdf
--Manfred
biject
> If you are interested in the bijective BWT, then you might also want
> to read my paper on the bijective sort transform.
>
>
> --Manfred
Is this the same paper in the same final form you have submitted
to Prague Stringology Conference 2009 or do you intend to make
changes.
Also since I have never been to Prague what do the people who
post here think of it as a conference. Is it a good one to go to or
what? I would really like to know. I may try to go I have to figure
out
just what I need to do.
Thank You
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"
Boon
> [...] I have to figure out just what I need to do.
Could you, perhaps, change your signature delimiter
from DASH DASH NEWLINE
to DASH DASH SPACE NEWLINE
mk
> >http://www.informatik.uni-stuttgart.de/fmi/ti/mitarbeiter/Kufleitner/...
> Is this the same paper in the same final form you have submitted
> to Prague Stringology Conference 2009 or do you intend to make
> changes.
Yes, this is the final (conference) version of the paper.
> Also since I have never been to Prague what do the people who
> post here think of it as a conference. Is it a good one to go to or
> what? I would really like to know. I may try to go I have to figure
> out just what I need to do.
PSC certainly is a rather small conference and its scope is all sorts
of what is called "combinatorics on words" which to some extend
includes (combinatorial) data compression algorithms.
You might not have to wear a tie ;-)
--Manfred