Re: QI and MQ Coder: First real-life experiences

Hi,

> Maybe I didn't communicate the central point well enough
> in the summary ([M1] or [N7] p.9 [T3]), but QI is an
> entirely different seed pattern, almost an antipode
> of the AC pattern on every key element, [M1].

Partially. QI is currently "nothing but" a smart way how to encode a
permutation of k LPS within n symbols. The question now is, how can
this be turned into a good model for something. While this is
particulary easy for AC and its coding methology, it seems
particularly hard for QI right now. For example, to make this a good
model, one could choose block sizes dynamically, but then would need
to find a smart way how to find block starts and ends that make sense
taking the side-channel (the number k) into account thus to allow
an overall coding gain.

The question is now how to find good models of audio or visual
data that match this QI coding paradigm. For AC, the model is
quite clear: Markov chains are mathematically well understood,
and there is a toolchain of mathematics how to deal with them.

Or to put my request in different terms:

Why is the QI coding paradigm a natural one for the mentioned
data sources?

It is not so that I don't want to understand that your beast has a
different approach. I just wanted - in my provocative nature - an
information how to make use of this approach for realistic data
sets, and the best way how to get it is to state a real contest.

> While much of the present coding and modeling is
> done within the AC's predictive and probabilistic
> pattern, there are only the isolated elements of
> the QI's modeling superstucture above, such as BW
> transform, along with the vastly rich substratum
> below, consisting of the highest quality mathematical
> results and algorithms of enumerative combinatorics,
> extremal and general finite set theory and other
> areas of finite/discrete math almost entirely
> untapped in this context.

BW is a pretty nice algorithm, but it is also pretty useless
for my field. (-; It doesn't fit to the data sources I care
about, and is a bad model for them. It might, or might not be
that this is about the same situation for QI. But then again,
this would contradicte your "is always better than AC" statement,
so what is it? (-:

So long,
Thomas
.

Relevant Pages