Decomposing tensors with structured matrix factors reduces to rank-1 approximations.
Pierre ComonMikael SørensenElias P. TsigaridasPublished in: ICASSP (2010)
Keyphrases
- singular values
- positive definite
- high order
- factors affecting
- nuclear norm minimization
- singular value decomposition
- order tensor
- factors influencing
- factors that influence
- structured data
- structure tensor
- diffusion tensor
- closed form
- real world
- key factors
- similarity matrix
- rank aggregation
- low rank
- tensor factorization
- symmetric positive definite
- symmetric matrices