Sparse and unique nonnegative matrix factorization through data preprocessing
From MaRDI portal
Abstract: Nonnegative matrix factorization (NMF) has become a very popular technique in machine learning because it automatically extracts meaningful features through a sparse and part-based representation. However, NMF has the drawback of being highly ill-posed, that is, there typically exist many different but equivalent factorizations. In this paper, we introduce a completely new way to obtaining more well-posed NMF problems whose solutions are sparser. Our technique is based on the preprocessing of the nonnegative input data matrix, and relies on the theory of M-matrices and the geometric interpretation of NMF. This approach provably leads to optimal and sparse solutions under the separability assumption of Donoho and Stodden (NIPS, 2003), and, for rank-three matrices, makes the number of exact factorizations finite. We illustrate the effectiveness of our technique on several image datasets.
Recommendations
- Non-negative matrix factorization with sparseness constraints
- Learning sparse representations by non-negative matrix factorization and sequential cone programming
- Semidefinite programming based preconditioning for more robust near-separable nonnegative matrix factorization
- Using underapproximations for sparse nonnegative matrix factorization
- Nonnegative matrix factorization: model, algorithms and applications
Cited in
(27)- Nonnegative matrix factorization requires irrationality
- Low permutation-rank matrices: structural properties and noisy completion
- Matrix-wise \(\ell_0\)-constrained sparse nonnegative least squares
- Quasi-orthogonalization for alternating non-negative tensor factorization
- Nonnegative matrix factorization via archetypal analysis
- Heuristics for exact nonnegative matrix factorization
- A nonlinear matrix decomposition for mining the zeros of sparse data
- Learning sparse representations by non-negative matrix factorization and sequential cone programming
- Methods for nonnegative matrix factorization based on low-rank cross approximations
- Common information and unique disjointness
- Efficient preconditioning for noisy separable nonnegative matrix factorization problems by successive projection based low-rank approximations
- Semidefinite programming based preconditioning for more robust near-separable nonnegative matrix factorization
- Nonnegative decomposition of functional count data
- Using underapproximations for sparse nonnegative matrix factorization
- A Blind Source Separation Technique for Document Restoration
- Lower bounds on nonnegative rank via nonnegative nuclear norms
- Uniqueness of Nonnegative Matrix Factorizations by Rigidity Theory
- Refinement of Hottopixx method for nonnegative matrix factorization under noisy separability
- On the construction of non-negative dimensionality reduction methods
- A survey on deep matrix factorizations
- SAGA: sparse and geometry-aware non-negative matrix factorization through non-linear local embedding
- Multiple linear regression models for random intervals: a set arithmetic approach
- Provably robust blind source separation of linear-quadratic near-separable mixtures
- Feature re-factorization-based data sparse representation
- Orthogonal joint sparse NMF for microarray data analysis
- Issues of stability and uniqueness of stochastic matrix factorization
- Neural nonnegative matrix factorization for hierarchical multilayer topic modeling
This page was built for publication: Sparse and unique nonnegative matrix factorization through data preprocessing
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5405228)
