Subject: Re: Improve the Calendar ?
From: rdadams@access2.digex.net (Dick Adams)
Date: 10 Jan 1997 19:12:15 -0500
Gary A Howard wrote:
> William H. Becker wrote in article
>> Wouldn't an improved international civil calendar
>> be a great boon in many sectors; scheduling, communications,
>> better statistical comparisons, budgeting, reduced confusion,
>> fixed day-date relationships, etc. etc.?
>> With the upcoming start of a new year, new century, and new
>> millennium, isn't this a good time to give this issue some
>> attention ? I sent U.S. Vice Pres. Gore info similar to that
>> covered in URL listed below and in Nov. 1993 he wrote me that the
>> idea of an improved calendar "deserves serious consideration".
>> ISN'T IT ABOUT TIME ? ?
>>
>> Suggest you look at ideas on Home Page for Calendar Reform at URL:
>> http://ecuvax.cis.ecu.edu/~pymccart/calendar-reform.html
>>
>> billbecker@msn.com
> I agree, but we can't even get Americans to move to the yy.mm.dd
> format from their early American m/d/y style. Heck, we can't even
> get them using the metric system!
Actually ccyy.mm.dd is the appropriate format. It is the preferred
format for anyone who needs to collate. However, the most common
international format is dd name-of-month ccyy -- and I changed to
that a few years ago.
But, I have yet to come across a compelling reason to convert to
metric. Most 19th century countries that converted to metric did
so at the point of a Napoleonic sword at their throat.
Dick
Subject: Re: Karhunen-Loeve transformation
From: "Jonathan G. Campbell"
Date: Sat, 11 Jan 1997 19:19:17 +0000
Bradley Jones wrote:
>
> Does anyone know what this is?
The (continuous) K-L (Karhunen-Loe\ve transform is related to the
Fourier transform in the sense that it yields a method by which a random
process (e.g. of time) x(t) to be expressed as a series expansion:
(1) x(t) = sum k=1 to inf { ak.Fk(t)}
The coefficients of the expansion are uncorrelated
(2) E{ai.aj} = 0, i!=j,
The Fk(.) are obtained as the eigenfunctions of the autocorrelation
function, R(t,s), of the random process, i.e. solutions of the integral
equation
(3) integral R(t,s).F(s).ds = l.F(t)
There are a family of solutions, eigenfunctions, Fk(t), k = 1...,
together with corresponding eigenvalues lk. The lk are analogs of the
Fourier spectrum, and the eigenfunctions are analogs of the
trigonometrical functions sine(.) and cosine(.).
It is highly likely that you are interested in the discrete
Karhunen-Loe\ve, in which case the summation in (1) is k=1 to N, where N
is the length of the sampled data sequence, and the integral in (3)
becomes a summation; R(t,s) becomes a correlation matrix (assuming
zero-mean data).
In this discrete case, I believe that K-L transform is equivalent to
principal components analysis, although I think that the exact
definition of PCA is something on which people differ. Moreover, this is
one of the many topics in which statisticians and electrical engineers
seem to use their own terminology, and, indeed, seem to go their
separate ways.
[above I use e\ to signify e-grave; a minority of authors wrongly spell
it using e-acute; I use Fk(t) for the normal phik(t), and lk for
lambdak]
The name Karhunen-Loe\ve (respectively, Finnish and French statistician)
and was probably introduced by [Davenport and Root, 1958].
The K-L expansion has certain optimal qualities: uncorrelatedness of the
coefficients, and rate of convergence of the series. These make it
interesting for data-compression, though there are practical
difficulties: e.g. generally no 'fast' algorithm.
However, for a certain class of two-dimensional image random processes,
the discrete cosine transform is a good approximation. Also, for
one-dimensional *stationary* random processes, it can be shown that the
Fourier transform is the (optimal) K-L transform.
[Rosenfeld & Kak, 1982] is a good source of a concise yet rigorous and
comprehensive treatment of K-L, though from the point of view of
two-dimensional random processes.
You can simple intuitive feel for K-L by considering a set of
n-dimensional data points; the (rotated) axis, upon which the data
projected gives the greatest variance, corresponds to the first
eigenvector of the correlation matrix, and that variance to the
corresponding eigenvalue; and so on, picking the next maximum axis that
is orthogonal to the earlier. Projected onto these axes, the data
obtained have a diagonal correlation matrix (see eqn. 2 above).
On another note, if you think of spinning a three-dimensional object
like you would a spinning-top, there will be an optimal axis about which
to spin it; about that axis, not only will the inertia (analog variance)
be greatest -- and it will spin longest -- but also the inertia matrix
(analog correlation matrix) will be diagonal -- end hence the object
will spin without wobbling.
Refs.
W.B. Davenport and W.L. Root, 1958, An Introduction to the Theory of
Random Signals and Noise, McGraw-Hill, reissued by IEEE Press, 1987.
A. Rosenfeld and A.C. Kak, 1982, Digital Picture Processing, Volume 1,
Academic Press.
Hope this helps, but, as I always say in forums like this, I apologise
for the nausea that I may have caused in any mathematicians!
Jon Campbell
--
Jonathan G. Campbell, ISC/ISE, University of Ulster, Magee College,
Derry, BT48 7JL, Northern Ireland. tel +44 1504 375367, fax 370040.
JG.Campbell@ulst.ac.uk http://www.iscm.ulst.ac.uk/~jon/
Subject: Combining Neural/Fuzzy Models with Statistical Models
From: agray@commerce.otago.ac.nz (Andrew Gray)
Date: Sun, 12 Jan 1997 02:00:06 GMT
Hi,
I'm working on combining neural networks and fuzzy logic models
with statistical techniques (regression and data reduction) for
software metrics (for example, predicting development time based on
the type and size of system). While there has been a lot of work on
neural-fuzzy, neural-genetic, fuzzy-genetic, etc. type systems I've
only ever found a small number of researchers using AI/statistical
techniques (presumably at least partially an indication of how few AI
researchers follow the statistical side of things, and vice versa).
If anyone out there is working on, or knows of some such work, then
I'd be really interested in hearing from you. If anyone would like to
post their ideas on this to the newsgroup, it would be nice to get
some discussion going.
Cheers,
Andrew
Software Metrics Research Laboratory, University of Otago
Phone: +64 3 479 5282 Fax: +64 3 479 8311
email: agray@commerce.otago.ac.nz
http://divcom.otago.ac.nz:800/COM/INFOSCI/SMRL/people/andrew/andrewg.htm (Home Page)
http://divcom.otago.ac.nz:800/COM/INFOSCI/SMRL/home.htm (SMRL's Home Page)
Subject: Beware the Woodside "Literary Agency"
From: Jack Mingo
Date: Sat, 11 Jan 1997 14:39:00 -0800
My apologies for the crosspost, but I'm trying to follow up on a
spammer/scammer who has posted 163 articles overnight. You may have
already seen warnings, but just in case:
The Woodside Literary Agency has recently posted ads in your
group. Beware of them--they are as legitimate as their forged e-mail
address.
People who have signed on with them have lost several hundred dollars,
and good number of their writerly dreams, to no good effect. There is no
evidence--and they've declined repeated opportunities to refute
this--that they've ever sold anything for any of their clients. In fact,
no editor I've asked has even heard of them.
Lately, they've been mailbombing one of their critics, and forging sex
messages with her phone number and address. You can see the whole sordid
story by searching on WOODSIDE or HITCHCOCK on Dejanews, or visiting the
misc.writing group. If you wish to complain about them, their real ISP
this week seems to be Earthlink (abuse@earthlink.net and
spam@earthlink.net).
Just for your information, if you are in the market for an agent, two
quick rules of thumb apply:
1. Legitimate agents rarely if ever advertise anywhere--most are too
busy turning away prospective clients to want to attract more.
2. No legitimate agent asks for money upfront. Real agents charge you
nothing until they've sold your writing, and then they take a smallish
percentage of the proceeds (10-15%).
Again, my apologies for busting in like this with a crosspost.
Jack Mingo
ambassador of goodwill
misc.writing
