| 1D vector pattern classification tips [message #5016] |
Tue, 12 September 1995 00:00 |
Russ Welti
Messages: 27
Registered: October 1994
|
Junior Member |
|
|
Just casting a line to see if anyone could direct me
to information (or pub domain software!) regarding
classification of 1D vectors (i.e. spectra, chromatograms)
into clusters.
In my case, I cannot know ahead of time what the
prototype example of each category will be, nor the
exact number of categories, but the data being
categorized is pretty simple: 1D vectors with
~l50 elements/pixels which represent preprocessed,
time-sampled intensity values from a DNA sequencer.
The plot looks like 3 to 10 distinct peaks with a nice
background level near zero.
The features here are also relatively straightforward:
number of peaks, their locations and intensities, etc.
I am told that cross_correlation is of limited value here,
due to the fact that one pattern can look just like another,
only offset by some unknown distance.
I am doing Web and bibliography searches on clustering
algorithms, with limited success. I'm afraid I need
something a bit below the level of the average thesis ;)
Has anyone done any of this before or have a good
pointer they could offer? A favorite textbook? A web page?
much thanks, as always,
/
Russ Welti /-\
(c-g)
University of Washington \-/
Molecular Biotechnology /
PO Box 352145 /-\
Seattle, WA 98195 (a-t)
rwelti@u.washington.edu \-/
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
