hcc_parallel {chickn} | R Documentation |
Compressed Hierarchical Clustering.
hcc_parallel( Data, W, K, maxLevel, ncores = 2, DIR_output = tempfile(), hybrid = FALSE, verbose = FALSE, ... )
Data |
A Filebacked Big Matrix n x N. Data signals are stored in the matrix columns. |
W |
A frequency matrix m x n with frequency vectros in rows. |
K |
Number of clusters at each call of the clustering algorithm. |
maxLevel |
Maximum number of hierarchical levels. |
ncores |
Number of cores. By default 4. |
DIR_output |
An output directory. |
hybrid |
logical parameter. If TRUE |
verbose |
logical that indicates whether dysplay the processing steps. |
... |
Additional arguments passed on to |
This function provides a divisive hierarchical implementation of COMPR
.
Parallel computations are performed using 'FORK' clusters (Linux-like platform) or 'PSOCK' clusters (Windows platform) using the parallel
package.
This function generates in the DIR_output
directory the following files:
'Cluster_assign_out.bk' is a Filebacked Big Matrix N x maxLevel
+1, which stores the cluster assignment at each hierarchical level.
'Centroids_out.bk' is a Filebacked Big Matrix with the resulting cluster centroids in columns.
The cluster assignment as a list of clusters with corresponding data vector indeces.
Keriven N, Tremblay N, Traonmilin Y, Gribonval R (2017). “Compressive K-means.” In 2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 6369–6373. IEEE.
data("UPS2") N = ncol(UPS2) n= nrow(UPS2) X_FBM = bigstatsr::FBM(init = UPS2, ncol=N, nrow = n)$save() K_W1 = Nystrom_kernel(Data = X_FBM, c = 14, l = 7, s = 5, max_neighbors = 3, ncores = 1, kernel = 'Gaussian')$K_W1 W = GenerateFrequencies(Data = K_W1, m = 20, N0 = ncol(X_FBM))$W C = hcc_parallel(Data = K_W1, W = W, K = 2, maxLevel = 4, DIR_output = tempfile(), ncores = 2)