14.5.2. k-Medoids Algorithms
In the k-means algorithm described in the previous section, each cluster is represented by the mean of its vectors. In the k-medoids methods, discussed in this section, each cluster is represented by a vector selected among the elements of X, and we will refer to it as the medoid. Apart from its medoid, each cluster contains all vectors in X that (a) are not used as medoids in other clusters and (b) lie closer to its medoid than to the medoids representing the other clusters. Let Θ be the set of medoids for all clusters. We will denote by IΘ the set of indices of the points in X that constitute Θ, and by IX − Θ the set of indices of the points that are not medoids. Thus, if for example Θ = {x1, x5, x13} is the set ...
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.
