fusion_cluster: Solve the fusion clustering problem with MCP penalization via...

In wenshuoliu/ncvxclustr: None-Convex Clustering Algorithm

Description

Solve the fusion clustering problem with MCP penalization via MM algorithm

Usage

fusion_cluster(X, U0, Phi, lambda, gamma, maxiter_mm, maxiter_cvx, tol, trace)

Arguments

X	the data, with the columns being units, the rows being features
U0	the initial guess of the centroid matrix U
Phi	the edge incidence matrix, defined as Phi_li = 1 if(l_1 == i); -1 if(l_2 == i); 0 otherwise
lambda, gamma	the parameters of MCP penalty function
tol	the duality gap tolerence
trace	whether save the primal and dual values of every iteration
maxiter	maximum iterations

Value

a list containing the solution U, V, and (optional) trace information

wenshuoliu/ncvxclustr documentation built on May 4, 2019, 5:21 a.m.