Some new results on orthogonally constrained Candecomp (Q649165)

Candecomp, developed by \textit{J. D. Carroll} and \textit{J. J. Chang} [Psychometrika 35, 283--319 (1970; Zbl 0202.19101)], is a method to minimize the residual sum of the squares in a scalar problem fitting problem derived from Indscal. In this paper, the authors first describe the Indort algorithm and the simultaneous update of the columns of relevant matrices. Then they present their main result by showing how to combine Indort with non-negative least squares to update saliences. This is followed by a section in which a successive approach, which finds one component at time, and compare its performance with the simultaneous approach.