topicmaps-comment message
[Date Prev]
| [Thread Prev]
| [Thread Next]
| [Date Next]
--
[Date Index]
| [Thread Index]
| [Elist Home]
Subject: [topicmaps-comment] Detecting Events in Computational Space
- From: paul <beadmaster@ontologystream.com>
- To: categoricalAbstraction@yahoogroups.com
- Date: Sat, 25 May 2002 11:05:29 -0400
Richard,
Please
look at the following paper, and make comments.
I feel
that you should be impressed with the notion that this work addresses the
perceptional act, and gives a synthetic knowledge system an ability to act and
see the response in real time in the real world. NIMA and others
should fund this soon, as it is an orginal and full new approach to better
National intelligence about THINGS THAT ARE REALLY
IMPORTANT.
I
could extend this paper so that stratified theory is more in the fore front; but
I realize that stratified theory has to be proven by building product... no
philosophical debate will win against a reductionist mainstream (as was shown in
the KMCI debate.) And the best way to extend this work is to simply read
Karl Pribram's 1991 Brain and Perception.
Pribram, K.H. (1971).
Languages of the Brain, experimental paradoxes and principles in
neuropsychology. New York: Wadsworth.
Pribram, K. H. (1991).
Brain and Perception: Holonomy and Structure in Figural Processing. Hillsdale,
NJ: Lawrence Erlbaum Associates.
Pribram, K. (Ed). (1993).
Rethinking Neural Networks: Quantum Fields and Biological Data. Hillsdale, NJ,
ERA
Pribram, K. (Ed). (1994).
Origins: Brain & Self Organization . Hillsdale, NJ,
Pribram, K. & King,
J. (Eds) (1996). Learning as Self-Organization. Mahwah, NJ, ERA
Pribram, Karl (1993) (Ed)
Rethinking Neural Networks: Quantum Fields and Biological Data, Hillsdale, NJ,
LEA
Pribram, Karl
(1994) (Ed). Origins: Brain & Self Organization. Hillsdale, NJ, LEA
Here
you will see that the notion of a convolution of a frame over data sources is in
his appendix. The convolution integrates the invariance of reponse in the
metabolic substrate of the brain regions, resulting in a phase coherence that is
then mediated by frontal lobe inhibition. (See my early work on
this:
Levine, D. & Prueitt, P.S. (1989.) Modeling
Some Effects of Frontal Lobe Damage - Novelty and Preservation, Neural Networks,
2, 103-116.
Levine D; Parks, R.; & Prueitt, P. S. (1993.)
Methodological and Theoretical Issues in Neural Network Models of Frontal
Cognitive Functions. International Journal of Neuroscience 72
209-233.
This
is now seen as being simple. My Russian friend Juri Kropotov has this
"architecture" in algorithms since 1997, as have many hundreds of others -
mostly from the Former Soviet Union. The architecture is stratified, as
seen by the two forms of a sum. 2+2 and 4 are the two levels.
Note that 4 = 1+3 is not correct in this sense, as the "cause" of 4 is 2 and a 2
added together. This meaning of the substructure is lost as the new
meaning of the whole is formed. (Damn it, I hope someone is following this
!!! ) The fact the 2 + 2 = 4 = 1+ 3 is an indication of the types of
mistakes that can be made in stratified semiotics if one is not willing look
beyond the types of formalism that created arthimetic. The cause of 4 is
lost, in exactly the same way as the cause of an image is lost if one uses an
ontology with only first order predicate logics. (We
claim.)
For
more on this stratified arithmetic see:
Finn, Victor (1991). Plausible
Inferences and Reliable Reasoning. Journal of Soviet Mathematics, Plenum Publ.
Cor. Vol. 56, N1 pp. 2201-2248
Finn, Victor (1995). JSM-reasoning for
control in open (+/-) worlds, in J. Albus, A. Meystel, D. Pospelov, and T
Reader, (Eds), Architectures for Semiotic Modeling and Situational Analysis in
Large Complex Systems, AdRem, Bala Cynwyd, PA
Finn, Victor (1996a). Plausible
Reasoning of JSM-type for Open Domains. In the proceedings of the Workshop on
Control Mechanisms for Complex Systems: Issues of Measurement and Semiotic
Analysis: 8-12 Dec. 1996
Finn, Victor (1996b) Basic concepts of
Quasi Axiomatic Theory, presented at the QAT Teleconference, New Mexico State
University and the Army Research Office, December 13, 1996.
and
http://www.ontologystream.com/IRRTest/Evaluation/ARLReport.htm
(In
physics and chemistry, one does not often have two ways of producing the same
non-organic compound. However, in biology we often have catolytic
micro-environments that will compose **functionally** equivelent organic
compounds.) Natural language has this same **complex** feature, when
one thing can have "response degeneracy" (I use here Gerald Edelman's
term).
Edelman, G. M. (1987). Neural Darwinism.
New York: Basic Books.
Peter
Kugler's work on the Rosen diagrams is able to formally define "complexity" with
a Rosen type diagram having this property. (Rosen himself may have never
made this definition explicit.)
Kugler's expertise is modeling "perceptual system" and the logic
interfaces; over the years his work
has involved the development of real-time modeling
interfaces--in the form of complete simulation environments. Over the past
two years this work has been consolidated
into some very elegant design principles
for perceptual systems that directly
address problems of observability and measurement in living systems. This work goes well beyond most notions of
perception and memory and is completely grounded in demonstration experiments, a modeling paradigm, and a metric
for semantic content.
Kugler , P.N. &
Turvey, M.T. (1987.) Information, natural law, and the self-assembly of rhythmic
movements. Hillsdale, NJ: LEA.
Kugler, P.N., Shaw,
R.E., Vincente, K.J. and Kinsella-Shaw, J. (1990). Inquiry into intentional
systems I: Issues in ecological physics. Psychological Research, 52: 98-
121.
As
part of the development of product, the paper describes how a cognitive graph
type system will serve as a repository of vetted knowledge artifacts derived
from invariance detecting convolutions. Shallow Link Analysis involves a
frame with four slots, filled by having two records (of two columns of
data) where one of the column values is repeated.
Mary
knows John, Mary knows Jim, maybe Jim and John know each other. The
convolution occurs as this frame is tested over each pair of records. The
combinatorics involved is by-passed using I-RIBs.
This
work has now been generalized to any frame with slots and fillers (May 2002,
Prueitt).
When
the convolution of the frame finds a pattern, the elements of the pattern are
encoded into a special tree structure (invented by Don Mitchell), if the pattern
has been found before a reference pointer is made. So one has a tree where
the leaves are each representing a pattern and a link list that encodes (using a
single small pointer) all occurance of this pattern. Even the "first"
occurance is encoded. We actually have a novel event detection capability
right out of the box. (with no learning regime required). There is no
boot-strap - and the system builds its own ontology without human intervention.
(We showed this with the notion of relative primes and the fundamental theorems
of SLIP technology. Of course, this went over the heads of the less
informed individuals in Industry - but it is quite simple and can be
demonstrated within an hour given any new data set.)
The
use of this architecture in COBOL software rehosting is being contemplated by
Don and I and Goren Osterman.
My new
work has generalized the shallow link analysis that I showed to Industry
last year.
The
general algorithm will be put into the Knowledge Operating System as a
measurement device so that a human can in real time act to change how the data
stream is being seen. This vision of a synthetic perception system
was assisted by Nathan Einwechter, Dean Rich, Jimmy Alderson, Martin Dion,
Nan Gelhard, Cameron Jones, Don Mitchell, Don Tobin and Scott Wimer.
(I
know that this architecture might seem difficult to understand at first, but
this is following Pribram's architecture for the human perceptional system, and
fullfills my interest in seeing a perceptional measurement system that looks
into the data stream.)
My
sense is that NIMA will have a number of Cognitive Graph systems to choose from
and that if I am to win this competition then I must work with whomever they
have choosen for the knowledge bases.
However, I am now just beginning to work on a NIST ATP proposal (Gate 1)
that needs to be addressed by June 10. We all need to have funding to bind
together the work of the group.. it just can not be done intellectually
only. The NIST proposal will be a consortium of groups that become bound
together by the project. We need 50% funding from the capitalization of
OSI (using 25% of our initial stock), and from revenues generated by the
BCNGroup and OSI.
I have
a number of innovations that I am trying to integrate into the "New Computer
Science", but in many cases the various integration works are encumbered by
non-disclosure agreements. It is good that you, at least, are being open
enough so that others see what the H--- you are talking about - in the
detail.
The
group in cA is a good group, and others can join if they ask
permission:
***
A note
to the group. I have made all necessary arrangements to begin the planned
$500,000 capitalization program (25,000 shares) by August 1st, 2002. We
are looking for a firm to handle this capitalization.
This
is a simple and legitimate first step in establishing OSI as a Corporation with
employees and profits.
We are
held by Rule 504 to investors that have incomes of more than 200K per year,
unless a detailed business plan is produced and a set of accounting procedures
are put in place. At this point we will be able to sell a share of stock
for 20 dollars.
Meanwhile, those who wish to join the work may become a member of the
BCNGroup:
***
Commerically, it is not acceptable to me to have a knowledge base that is
not the Mark 3. For the government funding proposals we are going to work
on standard APIs between the KOS string processor and knowledge bases.
This means an XML and Topic Maps team will be hired.
Paul:
I
have the last of my theory series letters in the stream to you. If we can post
the path files, then that letter can become a part of this group's
discussion.
Then, I can take up the EditForm-Builder Letters which are now almost 2
months, behind my scheduled plan. Those letters address the problem of a
collaborative group of editors with different source materials conspiring to
produce an integrated ontology that will alternately merge certain ideas or
build a contrasting world from competing views. The World Wide Web people and
Semantic Web enthusiasts were particularly interested in our evolving
methodology and experience with doing this. It is part of my tasks for
finalizing Mark 3's EditForm Grammar and fixing our Builder 3.0 requirements
in order to complete Mark 3's backend specifications.
This
whole conversation is on human markup -- past, present, and
future.
Do
you have some suggestions? Would you like to set any preliminaries or make
your comments as we go along? I would be very interested in using this
discussion to better understand the implications of your theories of
stratification in a collaborative framework of work products which I have
grown to know very well.
My
sense is that this group or, on your advice, a still wider circle might
be the proper center for those discussions which I think will go to
considerable depth.
Dick
Rex,
Before (or as) I
address some thoughts about primary schema (and more fully
speak to
Claude's very clear remarks) and the formation of ontology, I wish
to ask
if you can search your soul for an answer to what about the
biological
models of intelligence/perception/cognition/being is troublesome
for most
computer scientists. (Other than some of them are just very
wrong,
and poor science.) Is it merely a problem of sorting out the
proper science
from the mythology (such as humans are the only animal
that has a cognitive
process)?
I am sure I do not fully understand
why there is such bad blood between the
AI camp, for example, and the
notion to biology is relevant.
I do know that the biological model is
not well represented as a model that
computer science can address
(yet).... but what if a new model was advanced
that was biologically more
relevant and that was implementable as computer
science. I am of
course refering to my tri-level architecture work, but
only as a
summative work that hopes to bring the cognitive neuroscience of
Pribram,
Shaw and others into the e-commerce and intelligence
technologies.
Your thoughts?
Anyone
else?
To unsubscribe from this group, send
an email
to:
categoricalAbstraction-unsubscribe@yahoogroups.com
Your
use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
| Yahoo! Groups
Sponsor |
ADVERTISEMENT
;) | |
|
To
unsubscribe from this group, send an email
to:
categoricalAbstraction-unsubscribe@yahoogroups.com
Your
use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
[Date Prev]
| [Thread Prev]
| [Thread Next]
| [Date Next]
--
[Date Index]
| [Thread Index]
| [Elist Home]
Powered by eList eXpress LLC