Nonnegative 3-way tensor factorization via conjugate gradient with globally optimal stepsize.
Jean-Philip RoyerPierre ComonNadège Thirion-MoreauPublished in: ICASSP (2011)
Keyphrases
- tensor factorization
- globally optimal
- step size
- convergence rate
- faster convergence
- matrix factorization
- convergence speed
- negative matrix factorization
- collaborative filtering
- graph cuts
- cost function
- tag recommendation
- singular value decomposition
- nonnegative matrix factorization
- global optimum
- linear program
- maximum likelihood
- primal dual