Compression - 1stWorks Corporation Awarded Patent for Quantized Indexing

Author

1stWorks Corporation Awarded Patent for Quantized Indexing

markn@ieee.org

2007-01-09, 6:56 pm

http://biz.yahoo.com/bw/070109/20070109005831.html?.v=1

1stWorks Corporation Awarded Patent for Quantized Indexing, a
Fundamentally More Efficient and Faster Entropy Coding Algorithm for
General Purpose Data Compression.

Tuesday January 9, 10:49 am ET
1stWorks proven and patented data compression technology vital to all
data transfer applications, including audio, video, storage and search.

BOSTON--(BUSINESS WIRE)--1stWorks Corporation, a leader in real time
communications software and data compression technology, has been
granted a U.S. patent for its development of Quantized Indexing, a
faster and more efficient form of entropy coding.

Invented by 1stWorks Chief Scientist Ratko Tomic, Quantized Indexing is
the first practical form of data compression to apply the mathematical
discipline of enumerative combinatorics to entropy coding and modeling,
delivering demonstrably superior performance in both efficiency and
speed over the best arithmetic coder alternatives currently available.

[see link for rest]

So, not sure what "enumerative combinatorics" actually is, I turn to
Wikipedia and see:

"Combinatorial enumeration is a subfield of enumeration that deals with
the counting of objects whose symmetries do not exist or, if they
exist, are combinatorial in nature."

Which is close I guess. But I probably should buy the book:

http://www.amazon.com/Enumerative-C...y/dp/0521663512

|
| Mark Nelson - http://marknelson.us
|

Matt Mahoney

2007-01-10, 9:56 pm

markn@ieee.org wrote:
> http://biz.yahoo.com/bw/070109/20070109005831.html?.v=1
>
> 1stWorks Corporation Awarded Patent for Quantized Indexing, a
> Fundamentally More Efficient and Faster Entropy Coding Algorithm for
> General Purpose Data Compression.
>
> Tuesday January 9, 10:49 am ET
> 1stWorks proven and patented data compression technology vital to all
> data transfer applications, including audio, video, storage and search.
>
> BOSTON--(BUSINESS WIRE)--1stWorks Corporation, a leader in real time
> communications software and data compression technology, has been
> granted a U.S. patent for its development of Quantized Indexing, a
> faster and more efficient form of entropy coding.
>
> Invented by 1stWorks Chief Scientist Ratko Tomic, Quantized Indexing is
> the first practical form of data compression to apply the mathematical
> discipline of enumerative combinatorics to entropy coding and modeling,
> delivering demonstrably superior performance in both efficiency and
> speed over the best arithmetic coder alternatives currently available.
>
> [see link for rest]
>
> So, not sure what "enumerative combinatorics" actually is, I turn to
> Wikipedia and see:
>
> "Combinatorial enumeration is a subfield of enumeration that deals with
> the counting of objects whose symmetries do not exist or, if they
> exist, are combinatorial in nature."
>
> Which is close I guess. But I probably should buy the book:
>
> http://www.amazon.com/Enumerative-C...y/dp/0521663512
>
> |
> | Mark Nelson - http://marknelson.us
> |

Or read the papers linked from the article.
http://www.1stworks.com/ref/qi.htm

Or the very long thread on this topic from about a year ago begun by
nightlight, who is the author of the papers and the patent.
http://groups.google.com/group/comp...99934202277c192

The dispute was over whether quantized indexing (aka enumerative
coding) is faster than arithmetic coding (possibly) and codes more
compactly (not significantly). The di

vantage is that QI requires
independent, identically distributed bits. It does not have the
flexibility of arithmetic coding to compress each bit with a different
probability.

-- Matt Mahoney

Sponsored Links