paleonet Systematics surprisingly interesting Pt.I

[John Jackson <strangetruther@xxxxxxxxx>]

Some years ago, I proposed, if put in charge of palaeo
at NHM London, to introduce it to the wonderful world
of knowledge engineering.  I didn't get the job, but I
knew the process was in train anyway, had been for
some time and was bound to continue. 

The Systematics Association's symposium "Algorithmic
Approaches to the Identification Problem in
Systematics" on Fri 19th Aug in NHM dealt largely with
the use of automated systems to identify the species
of individuals, sometimes in the field.  Artificial
neural nets abounded; DAISY was mentioned again. 
Norman MacLeod (who got my job) gave a well presented
account of his use of an ANN to identify plankton. 
Marvin Minsky's spirit (though he's still very much
alive) was strongly present that day - not only did he
co-invent the term artificial intelligence and
implement the first ANN, and then gave a permanent
emotional edge to the term 'perceptron' (a form of
which NMacL used) but he even invented the confocal
scanning microscope used in that project to build 3D
models of specimens.

That was a good and encouraging presentation; the only
problem was with the laser pointer: if by chance any
of the audience are trying to read the text, an
over-enthusiastic laser dot flicking around the screen
like a demented mosquito can effectively prevent it.  

The following week, at the continuation of the 5th
Biennial meeting of the Systematics Association at
Cardiff uni, Mark Wilkinson, Mike Benton's first PhD
student, and an excellent speaker, had the good
fortune to have his laser pointer die on him, and was
able to demonstrate the more dignified and effective
rhythm of a simple bamboo pole.

Cardiff was a 3-4 day meet and I hoped it would
feature the impact of AI with the genome project,
crunching traditional simplistic cladistics in
between.  Maybe it did; if so the sound was of MW's
careful delivery smoothly encouraging a more
open-minded approach (in lieu of exactly what, I had
to guess) being echoed by speaker after speaker. 
Davide Pisani had worked on "the genome project" and
applied methods of a theme of the meet - compatibility
analysis, nicely outlined by MW - to genomics, using
in particular split-fit.  In a way it was a relief to
see an account of the history of techniques some of
which I had re-invented myself, eg the principle at
least of compatibility analysis.  All that remains is
to run simulated evolution containing all types of
patterns of eg homoplasy, and use neural nets to
identify which phylogenetic tools identify which
patterns, and which tools correct for them... and of
course defy the old culture and use these methods
effectively on dino-bird data, and then Tell The
Public!  Opening those Pandoras boxes that need it is
an essential duty for a scientist.

The public are dying to know the answer to the riddle
of the turtle - at least I am.  Simon Harris from the
uni of Newcastle upon Tyne said molecular results
tended to point to diapsid, but morphological tended
to point elsewhere, and he progressively removed the
more obstreperous morph chars - using "boil-down" -
and showed some convincing graphs that homed in on a
diapsid origin even with morph. data.  Good - the
heuristic: "if it's got odd patterns of dermal things
on its back, guess archosaur, guess originally aerial"
gets more weight, at least in the first half.

Offering further alternatives to simplistic cladistics
was Peter Wagner.  Always a delight to hear, despite
his fast delivery and thick mathematical accent (it's
clearly his first language) he combines diamond sound
philosophy of science with a masterly blend of theory
and practice.  Just knowing that tells you he's not a
slave to fashion, though you could guess it from his
waist-coat too.  This time he combined compatibility
analysis with his regular theme of analysing rates of
change in different branches.  His lesson has still
not been learned that claiming to rely on parsimony
but not realising variable rates of character change
in various branches is a source of complexity too
(which has to be included in any measure of overall
simplicity) means your approach is as unjustified
theoretically as it is useless practically.

And Then There Was THIS...
(needs broadband)
www.4threvolt.com/EEMovie.html .

(contd. on next posting.)

JJ

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com