Quick start:

- extract files to directory (which You probably did already),
- start MATLAB / GNU Octave >= V 2.1.50,
- cd to target directory,
- on prompt type> embed_acidose
- for GNU Octave type> replot

A scatter plot display should have appeared.
11-dimensional data points, 20 of which, are embedded in 2D.

Redo embed_acidose for some times to get best solution.

The best solution has got maximum r; for acidose data r ~ 0.943.

Further actions:
- Uncomment in hitmds_hybrid.m, hitmds.m line: plot(Y(:,1),Y(:,2),'*')
- In embed_acidose play around with parameters and data similarity measure.
- See hitmds_hybrid.m to compare standard implementation and vectorized code.


Good luck
Marc Strickert

(stricker@ipk-gatersleben.de)
(Tue Jun  5 15:54:35    2007)

* last modification: new: linear learning rate decay after 1/2 max_iter
// EOF