fit_continuous: Fitting the Tucker model to a tensor

In tensorordinal: Tensor Noise Reduction and Completion Based on Ordinal Observations

Description

Fit the Tucker model to a tensor with possibly missing values

Usage

fit_continuous(ttnsr,r,alpha = TRUE)

Arguments

ttnsr	an observed tensor
r	a rank to be fitted (Tucker rank)
alpha	a signal level alpha = TRUE if the signal level is unknown

Value

a list containing the following:

C - an estimated core tensor

A - estimated factor matrices

iteration - the number of iterations

cost - log-likelihood value at each iteration

Examples

# Latent parameters
library(rTensor)
alpha = 10
A_1 = matrix(runif(15*2,min=-1,max=1),nrow = 15)
A_2 = matrix(runif(15*2,min=-1,max=1),nrow = 15)
A_3 = matrix(runif(15*2,min=-1,max=1),nrow = 15)
C = as.tensor(array(runif(2^3,min=-1,max=1),dim = c(2,2,2)))
theta = ttm(ttm(ttm(C,A_1,1),A_2,2),A_3,3)@data
theta = alpha*theta/max(abs(theta))
adj = mean(theta)
theta = theta-adj
omega = c(-0.2,0.2)+adj

# Observed tensor
ttnsr <- realization(theta,omega)@data

# Estimation of parameters
continuous_est = fit_continuous(ttnsr,c(2,2,2),alpha = 10)

tensorordinal documentation built on Jan. 13, 2021, 9:17 p.m.