Kalman Filters

Hans Blom
Would you please tell us illiterates what a Kolman Filter is! I tried and
tried on Web Search and could not get a single hint. CSGNet is an
interdisciplinary group. Common courtesy would require at least a hint of
what someone in another specialty is talking about. I hope this is not what
is going to happen at Flagstaff. I have gone to a lot of effort to make it
to the conference. It was not easy. Please don't disappoint me. If all I am
going to hear is a babble of electronic engineers discussing filagree of
circuits, I will tear out my few remaining hairs.
I find the writings of Bill Powers very considerate in terms of addressing
the various disciplines involved in control theory. I wish the same were
true of some of our techies.
PCT is a clear concept that needs still further clarification. I am in the
field of psychology and find PCT suddenly straightens out much of the
confusion in the field. If we are to find acceptance, we will have to cut
out the hair splitting. It is no sin to simplify!
Would it be too painful to include a few words of clarification with some
of the highly technical posts?
Ellery

... I hope this is not what is going to happen at Flagstaff. I have gone to

a >lot of effort to make it to the conference. It was not easy. ...

Fear not! Most of the folks at the conference will not put up with either
"undefined" jargon or "in" terminology.
-bill
b.leach@worldnet.att.org
ars: kb7lx

[Hans Blom, 960710]

From: Ellery Lanier <elanier@CRL.NMSU.EDU>
Subject: Kalman Filters

Would you please tell us illiterates what a Kolman Filter is! I tried and
tried on Web Search and could not get a single hint. CSGNet is an
interdisciplinary group. Common courtesy would require at least a hint of
what someone in another specialty is talking about.

One of my most recent posts may have been helpful. Kalman Filtering
is about establishing knowledge out of uncertain bits of evidence. It
is very similar to the ANOVA methods that are used in psychology to
analyze data. The main difference is that you don't process the data
when all measurements have been completed, but that you process
continuously. The end result is the same, but what is now extra, is
that you have intermediate results as well. The accuracy improves as
you collect more data. HOW this is done is very technical; that
cannot be helped. Is WHAT is done now somewhat clear?

I hope this is not what is going to happen at Flagstaff.

Rest assured: I won't be there.

I find the writings of Bill Powers very considerate in terms of addressing
the various disciplines involved in control theory. I wish the same were
true of some of our techies.

Techies tend to be somewhat technical ;-). What I wanted to show is
that, inside the computation, the ACCURACY of the result becomes
available, essentially for free. Not only can we say, for instance,
that a speed is 75 mph, but we can also say that we know with 99%
certainty that the speed is 75 +/- 3 mph. That may be important,
sometimes, e.g. when the speed limit is 70 mph. The accuracy is the
ubiquitous "p < 0.005" of the psychological literature, without which
nobody will believe you.

PCT is a clear concept that needs still further clarification.

I try to do that in my way. Sometimes you need to rigorously develop
concepts first, before you can cast them into "clear" language.

I am in the field of psychology and find PCT suddenly straightens
out much of the confusion in the field. If we are to find
acceptance, we will have to cut out the hair splitting. It is no
sin to simplify!

That is right. On the other hand, a simple explanation ONLY may
remain so fuzzy that it is useless.

Would it be too painful to include a few words of clarification with some
of the highly technical posts?

Ellery, I'll try, but sometimes this just isn't possible. When I give
a too simple explanation (as I've often done in the past), it may be
misunderstood and many unwarranted objections may be the result. In
such cases, detail is necessary. I'm sure that Bill agrees with this.

But I'm also sure that much else makes this list worthwhile to you,
eh?

Greetings,

Hans