Göm menyn

Fredag 4 november 2016, Göran Bergqvist, MAI

Talare: Göran Bergqvist, MAI

Titel: The higher-order singular value decomposition

Tid och plats: Fredag 4 november 2016, Hopningspunkten, kl. 10.15

Sammanfattning: The singular value decomposition of a matrix is a standard tool in engineering for low-rank approximation and compression of data. When data is given as a multidimensional array or tensor, several generalisations are possible. The most used today in real applications is the so-called higher-order or multilinear singular value decomposition presented by De Lathauwer, De Moor and Vandewalle in 2000. We describe this decomposition and some of its properties regarding computation, low-rank approximation and applications.


Sidansvarig: karin.johansson@liu.se
Senast uppdaterad: 2016-10-26