Re: Statistical Pattern Recognition

On Sat, 29 Dec 2007 12:21:02 +1000, j.wilkins1@xxxxxxxxx (John
Wilkins) wrote:

r norman <r_s_norman@xxxxxxxxxxxx> wrote:

On Sat, 29 Dec 2007 02:18:26 +1000, j.wilkins1@xxxxxxxxx (John
Wilkins) wrote:

r norman <r_s_norman@xxxxxxxxxxxx> wrote:

On Fri, 28 Dec 2007 15:33:47 +1000, j.wilkins1@xxxxxxxxx (John
Wilkins) wrote:

r norman <r_s_norman@xxxxxxxxxxxx> wrote:

<snips>

And, while I am at it, the thermostat does indeed perform a
calculation deriving a digital "error" signal from the current
temperature and the set point.

Please expand on that. Do you mean digital thermostats, or bimetallic
ones? It rather sounds like you think two strips of metal can do a
calculation of the Fisher information via a Cramer-Rao Bound, which is a
scientific miracle if anything is.

The mechanism within the thermostat is irrelevant; a digital computer
can be built from gears and cams and the like, relays, vacuum tubes,
transistors, or massively integrated circuits and still be a computer.

The important aspect is that the thermostat store, in analog or
digital fashion, a value called the 'set point', measures the ambient
temperature, and produces a logical signal (true/false valued) which
is the result of the computation: the current temperature is less than
the set point". Actually, thermostats including those based on a
bimetallic strip perform far more sophisticated computations
incorporating hysteresis (the actual set point changes depending on
the history of temperature change) and time (the current state must
exist for a predetermined duration). Rather a lot of engineering
design goes into the production of a thermostat.

Note: the bimetallic strip does NOT perform that calculation: it
must be part of a system (aka "thermostat") to do that. You see, the
computation emerges from the system organization!

So long as the system described includes the observer, in which case
Fisher information falls right out of it.

Are you saying that the temperature regulation in my house goes all to
hell if I am not there to watch it?

No, I am saying that if you are not there to observe it, the thermostat
has no information, and does no computation.

Yes, I know you claim that "performing a computation" is merely a
convention that I, as an observer analyzing the system, impose on what
the system actually does. I claim that if it walks like a duck and
...
... needs to be wound up, and it painted and made of wood, it is *not* a
duck.


And, to be serious just for a moment, I don't see where Fisher
information has anything at all to do with it. That is a purely
statistical measure and the thermostat system can be treated as a
purely deterministic system. There is no probability distribution for
which a variance can be computed.

Really? I thought that any measurement device would return a number of
values that gave you the expectation curve that the measurement was
accurate.

We are discussing gedanken thermostats and gedanken computers (i.e.,
we are mentally computing what such devices are and how they behave).
The thermostat simply measures the temperature and it gets the value
correctly. Period. The computation does not depend on the
statistical significance of the measurement. If the result of the
measurement is incorrect, the result of the computation is
correspondingly incorrect. It still is a computation.

.

Relevant Pages

  • Re: Statistical Pattern Recognition
    ... calculation deriving a digital "error" signal from the current ... temperature and the set point. ... The mechanism within the thermostat is irrelevant; ... Fisher information falls right out of it. ...
    (talk.origins)
  • Re: Statistical Pattern Recognition
    ... Wilkins) wrote: ... calculation deriving a digital "error" signal from the current ... temperature and the set point. ... The mechanism within the thermostat is irrelevant; ...
    (talk.origins)
  • Re: Statistical Pattern Recognition
    ... calculation deriving a digital "error" signal from the current ... temperature and the set point. ... The mechanism within the thermostat is irrelevant; ... Note: the bimetallic strip does NOT perform that calculation: it ...
    (talk.origins)
  • Re: Statistical Pattern Recognition
    ... calculation deriving a digital "error" signal from the current ... temperature and the set point. ... The mechanism within the thermostat is irrelevant; ... Note: the bimetallic strip does NOT perform that calculation: it ...
    (talk.origins)
  • Re: OT Wall street occupation.
    ... heat does not stay put. ... thermostat is set at that Hot House Harry optimum, ... be a five or ten degree swing in temperature in that particular room. ... thermostat will turn the room heating down. ...
    (alt.home.repair)