ldakmeans

LDA-K-means algorithm
[IDX,U,FVAL,C,ITER] = LDAKMEANS(X,K,VARARGIN)

 [IDX,U,FVAL,C,ITER] = ldaKmeans(X,K) produces a flat partitioning of the N-by-D
 data matrix (X) into (K) clusters. The algorithm iteratively performs Linear
 Discriminant Analysis and K-means clustering.

 [IDX,U,FVAL,C,FLAG] = ldaKmeans(X,K) returns the cluster assignment, (IDX);
 the projection matrix (U) used to perform dimensionality reduction; the value
 of the projection index (FVAL); the cluster centroids (C); and finally the
 iteration at which the algorithm terminated, (ITER). If LDAKMEANS fails
 to converge ITER=0
 
 [IDX,U,FVAL,C,ITER] = ldaKmeans(X, K, 'PARAM1',val1, 'PARAM2',val2, ...)
 specifies optional parameters in the form of Name,Value pairs. These are only
 used for visualisation

  'maxit' - Number of iterations to perform (default: 50)

  'ftol' - Tolerance level (default: 1.e-5)

  'verb' - Verbosity. Values greater than 0 enable visualisation during execution
    Enabling this option slows down the algorithm considerably
    (default: 0)

  'labels' - true cluster labels. Specifying these enables the computation of performance over 
    successive iterations and a better visualisation of how clusters are split

  'colours' - Matrix containing colour specification for observations in each of the (K) clusters

Reference:
C. Ding and T. Li. Adaptive dimension reduction using discriminant analysis and k-means clustering. 
Proceedings of the 24th International Conference on Machine Learning, pages 521-528, 2007.