A tightest convex envelope heuristic to row sparse and rank one matrices.
Alireza AghasiSohail BahmaniJustin K. RombergPublished in: GlobalSIP (2013)
Keyphrases
- nuclear norm
- low rank matrix
- low rank
- data matrix
- binary matrices
- matrix completion
- low rank matrices
- singular value decomposition
- matrix factorization
- low rank approximation
- convex optimization
- sparse matrix
- singular values
- rows and columns
- missing data
- coefficient matrix
- binary matrix
- positive semidefinite matrices
- frobenius norm
- rank minimization
- search algorithm
- optimal solution
- singular vectors
- missing values
- positive semidefinite
- linear combination
- original data
- kernel matrix
- discrete tomography
- high dimensional
- semi supervised
- least squares
- tabu search
- globally optimal
- convex relaxation
- sparse data
- convex sets
- simulated annealing
- dimensionality reduction
- combinatorial optimization
- collaborative filtering
- state space
- recommender systems
- nonnegative matrix factorization