R help - Dec 2008 - How to cluster the elements of a binary vector

I would appreciate some suggestions with clustering the rows of a binary matrix.
Basically given a binary vector (a generic row from matrix Bb) which contains an
unknown number of 1s mixed with 0s, the problem is to define clusters according
to the vague and incomplete description given in a paper:.
 " Determine the clusters  U1 ,U2 ,....,U c  for each scale (matrix  row)
in the binary matrix Bb separately. The clustering is performed scale by scale
without overlapping across the scales (rows) such that each cluster contains 0s
and a 1 positioned near the center of the cluster. If the whole row of the
matrix Bb doesn't have any 1s, then all coefficients in that scale are
assumed to form just one cluster. The end result of all this is a cluster
template that is used for feature extraction in .."
They ignore cases like:
1. a row of matrix Be containing a 1 on either end (first and/or last row
element)
2. a row of matrix Be containing two or more adjacent 1s (no 0 in-between)

Thak you  so much in advance,
Maura




tutti i telefonini TIM!


	[[alternative HTML version deleted]]