You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Xiangyong Cao, Yang Chen, Qian Zhao, Deyu Meng, Yao Wang, Dong Wang and Zongben Xu,
Low-Rank Matrix Factorization Under General Mixture Noise Distributions,
15th International Conference on Computer Vision (ICCV), Chile, Dec. 2015 (Oral)
[label,model,TW,OutU,OutV,llh,llh_BIC,p] = EM_PMoEP(InW,InX,r,param,p,lambda)
%Input:
InW: d x n x param.k indicator matrices
InX: d x n input data matrix
r: the rank
param:
--param.maxiter: maximal iteration number
--param.OriX: ground truth matrix
--param.InU: initialized factorized matrice U
--param.InV: initialized factorized matrice V
--param.k: the number of mixture components
--param.display: display the iterative process
--param.tol: the tolerance for stop
p: the candidate components
lambda: the tuning parameter
%Output:
label: the labels of the noises
model: model.eta, the precisions of the different EPs
model.Pi,the mixing coefficients
W: d x n weighted matrix
OutU: the final factorized matrix U
OutV: the final factorized matrix V
llh: the log likelihood
llh_BIC: the log likelihood used in BIC criterion
p: the selected components
=========================================================================
If you have any quesion, please contact Xiangyong Cao(caoxiangyong45@gmail.com)
About
This is the implementation of "Low rank matrix factorization under general mixture noise distributions."
