15.6. Valley-Seeking Clustering Algorithms
The method discussed here is in the same spirit as that of the previous section. Let p(x) be the density function describing the distribution of the vectors in X. An alternative way to attack the clustering problem is to view the clusters as peaks of p(x) separated by valleys. Inspired by this consideration, one can search to identify such valleys, and try to move and place the borders of the clusters in these valleys.
In the sequel, we discuss an iterative and computationally effective algorithm based on this idea [Fuku 90]. Once more, let V(x) be the local region of x, that is,(15.38)
where a is a ...
Get Pattern Recognition, 4th Edition now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
