Statistical Pattern Recognition

On Dec 22, 5:37 pm, "R. Baldwin" <res0k...@xxxxxxxxxxxxxxxxxxxx>
wrote:
"Seanpit" <seanpitnos...@xxxxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message

news:a1a7d913-cf2c-45eb-8848-4c88f33f35e4@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx





On Dec 22, 10:30 am, richardalanforr...@xxxxxxxxxxxxxx wrote:
On Dec 22, 7:44 am, eerok <krke...@xxxxxxxxxxxx> wrote:

Seanpit wrote:

[...]

Useful scientific conclusions based on valid statistical
analysis need not be published or even written down.

I for one wouldn't mind a bit if you explored this "not
writing down" concept more fully.

--
"It is not enough to conquer; one must learn to seduce."
-- Voltaire

Mind you, I'm rather taken by the idea of rats carrying out
statistical analyses. I wonder if Sean has access to a breed of
superintelligent rodents. But then he insists that we all carry out
such analyses all the time *without even knowing it*!

You don't have to be superintelligent to carry out a useful level of
statistical analysis.  All you have to do is be able to remember past
events. Even rats can do this.  If the location of the cheese in the
maze is generally in a particular place, even if that place is not
always the location of the cheese, it has been shown that rats can
figure out that it saves time, most of the time, to check this
location first.  That, my friend, is a valid use of statistical
analysis.  Statistically, the odds favor the appearance of the cheese
in a specific location - - and the rats were in fact able to figure
this out without writing anything down or consulting any other rats.

You are also able to do the same thing.  That is why you know that a
highly symmetrical polished granite cube is in fact "anomalous" from
the perspective of non-deliberate origin.  In fact, it is so anomalous
as to be clearly artifactual for most people - to include most
scientists (even if they were on Mars or some other alien planet).

The same thing is also true for your tool marks.  You don't need to
write down your past experiences with tools and tool marks to be able
to recognize the markings of a particular tool as being anomalous or
"artifactual" right away.  You are able to do this because your memory
is good enough at remembering past events and adding up the numbers
that you don't need to make a written record of them before you can
use them to help you make useful predictions.

That, again, is a valid use of statistics.  No one ever said that
statistics could only be done with pen and paper.  Very useful
statistics can be done entirely in ones own head.  Arguing otherwise
means that you don't understand the basic nature of learning itself.
Basic learning is based on statistics.

Sean, you are *guessing* that this is how cognition works.

It isn't just a blind guess. It is based on attempts to model what the
brain does with computers using neural-like networks, etc.

Where, from all
the recent brain research is the *evidence* that shows brains perform
statistical analysis during pattern recognition? How do you *know* what the
algorithm is?

We don't know for sure what algorithm the brain actually uses. But,
the best evidence availble strongly suggests very complicated
algorithms and statistical analyses are indeed in play. Learning isn't
just some magical processes. Rather, it seems to be mathemetically and
statistically based.


"By the time they are five years old, most children can recognize
digits and letters. Small characters, large characters, handwritten,
machine printed, or rotated all are easily recognized by the young.
The characters may be written on a cluttered background, on crumpled
paper or may even be partially occluded. We take this ability for
granted until we face the task of teaching a machine how to do the
same. . . Watanabe [163] defines a pattern as opposite of chaos . . .
Statistical pattern recognition has been used successfully to
design a number of commercial recognition systems. In statistical
pattern recognition, a pattern is represented by a set of d features,
or attributes, viewed as a d-dimensional feature vector. Well-known
concepts from statistical decision theory are utilized to establish
decision boundaries between pattern classes.

Anil K. Jain, Fellow, IEEE, Robert P.W. Duin, and Jianchang Mao,
Senior Member, IEEE, "Statistical Pattern Recognition: A Review", IEEE
TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, VOL. 22,
NO. 1, JANUARY 2000

http://www.mts.jhu.edu/~priebe/COURSES/FALL2003/550.730/jdm00.pdf

See also:
"A Statistical Learning/Pattern Recognition Glossary"

http://research.microsoft.com/~minka/statlearn/glossary/

And this:

"Statistical pattern recognition is based on statistical
characterizations of patterns, assuming that the patterns are
generated by a probabilistic system."

http://en.wikipedia.org/wiki/Pattern_recognition

Sean Pitman
www.DetectingDesign.com

.

Relevant Pages

  • Re: Statistical Pattern Recognition
    ... the recent brain research is the *evidence* that shows brains perform ... statistical analysis during pattern recognition? ... that it is an *associative memory* process, not statistical analysis. ... I've already provided several references dealing with the statistics ...
    (talk.origins)
  • Re: Statistical Pattern Recognition
    ... the way pattern recognition works in humans which concludes that it is ... the product of unconscious statistical analysis. ... this is called inferential statistics. ... You need to read the references more closely as well, ...
    (talk.origins)
  • Re: Statistical Pattern Recognition
    ... the way pattern recognition works in humans which concludes that it is ... the product of unconscious statistical analysis. ... behind pattern recognition. ... Statistics are not involved in all of the ...
    (talk.origins)
  • Re: Statistical Pattern Recognition
    ... the way pattern recognition works in humans which concludes that it is ... the product of unconscious statistical analysis. ... behind pattern recognition. ... That requires statistics. ...
    (talk.origins)
  • Re: God=G_uv proves 40k B.C. Creation
    ... The brain *IS* a "Box". ... >> feed a statistics hobbyist like you statistics info for your own private ... math, investigating the theory, or looking for scientific truth. ...
    (sci.physics.relativity)