DiMaI - Dipartimento di Matematica e Informatica "Ulisse Dini"

Carmeliza Navasca (University of Alabama)
terrà un seminario

Giovedì 27 Giugno alle 12.00 in sala conferenze Tricerri su

Numerical Methods for a Minimum Sum of Rank-one Tensor Decomposition

Abstract: We present some recent methods for computing a canonical polyadic
decomposition.
The Alternating Least-Squares (ALS), the most well-known technique, can be
inefficient with long "swamps" or may not work in finding tensor
decompositions. In this talk, we describe some new algorithms: an
optimization of the reduced least-squares functional and two variants of
the ALS for fully and partially symmetric tensors.