bttdSoftImpute: bttdSoftImpute

In namgillee/BTTSoftImpute: R Software for Block Tensor Train Decomposition for Missing Data Estimation

Description

bttdSoftImpute implements block tensor train decomposition for missing data estimation. It extends the SoftImpute method for matrix data imputation to higher-order tensor data.

Usage

bttdSoftImpute(Y, lambda = 0, rank_Y = 2, ttrank_max = 100,
  tol_dx = 1e-05, tol_df = 1e-05, tol_f = 0, maxit = 100,
  verbose = FALSE, X0 = NULL, method = "ALS")

Arguments

Y	A sparse matrix format with attributes $i, $j, $v, $nrow, $ncol, and $dimnames, which are [nnz]-vector, [nnz x N]-matrix, [nnz]-vector, scalar, [N]-vector, and list, respectively.
rank_Y	The rank of the estimated low rank matrix, X.
ttrank_max	The maximal TT rank of right singular vectors, X$v.

Details

Last modified: 2018.01.15. by Namgil Lee (Kangwon National University)

Value

X = list(u, d, v), v is in block TT format

namgillee/BTTSoftImpute documentation built on July 2, 2019, 8:35 p.m.