Successive Nonnegative Projection Algorithm for Robust Nonnegative Blind Source Separation

Abstract

In this paper, we propose a new fast and robust recursive algorithm for near-separable nonnegative matrix factorization, a particular nonnegative blind source separation problem. This algorithm, which we refer to as the successive nonnegative projection algorithm (SNPA), is closely related to the popular successive projection algorithm (SPA), but takes advantage of the nonnegativity constraint in the decomposition. We prove that SNPA is more robust than SPA and can be applied to a broader class of nonnegative matrices. This is illustrated on some synthetic data sets, and on a real-world hyperspectral image.

31 pages, 7 figures, 4 tables. Main changes: new numerical experiments on column-rank-deficient matrices, typos corrected, discussion on the comparison with XRAY

References in corpus (11)

• Fast and Robust Recursive Algorithms for Separable Nonnegative Matrix Factorization
• The Why and How of Nonnegative Matrix Factorization
• Fast Conical Hull Algorithms for Near-separable Non-negative Matrix Factorization
• Factoring nonnegative matrices with linear programs
• Sparse and Unique Nonnegative Matrix Factorization Through Data Preprocessing
• Robust Near-Separable Nonnegative Matrix Factorization Using Linear Optimization
• Learning Topic Models - Going beyond SVD
• Semidefinite Programming Based Preconditioning for More Robust Near-Separable Nonnegative Matrix Factorization
• Topic Discovery through Data Dependent and Random Projections
• Robustness Analysis of Hottopixx, a Linear Programming Model for Factoring Nonnegative Matrices
• Ellipsoidal Rounding for Nonnegative Matrix Factorization Under Noisy Separability

Cited by in corpus (28)

• Nonnegative Matrix Factorization for Signal and Data Analytics: Identifiability, Algorithms, and Applications
• Deep matrix factorizations
• A Critical Survey of Deconvolution Methods for Separating cell-types in Complex Tissues
• Algorithms and Comparisons of Non-negative Matrix Factorization with Volume Regularization for Hyperspectral Unmixing
• Generalized Separable Nonnegative Matrix Factorization
• Introduction to Nonnegative Matrix Factorization
• A Fast Gradient Method for Nonnegative Sparse Regression with Self Dictionary
• A consistent and flexible framework for deep matrix factorizations
• Using separable non-negative matrix factorization techniques for the analysis of time-resolved Raman spectra
• Dictionary-based Tensor Canonical Polyadic Decomposition
• Simplex-Structured Matrix Factorization: Sparsity-based Identifiability and Provably Correct Algorithms
• Enhancing Pure-Pixel Identification Performance via Preconditioning
• Near-Convex Archetypal Analysis
• Mixed Membership Graph Clustering via Systematic Edge Query
• Spectral Unmixing with Multiple Dictionaries
• Smoothed Separable Nonnegative Matrix Factorization
• Provably robust blind source separation of linear-quadratic near-separable mixtures
• Convex Programming Based Spectral Clustering
• Refinement of Hottopixx Method for Nonnegative Matrix Factorization Under Noisy Separability
• Greedy Algorithms for Cone Constrained Optimization with Convergence Guarantees
• Orthogonal Nonnegative Matrix Factorization with the Kullback-Leibler divergence
• Deep NMF Topic Modeling
• Dual Simplex Volume Maximization for Simplex-Structured Matrix Factorization
• Matrix-wise ${\ell }_0$-constrained Sparse Nonnegative Least Squares
• On the Robustness of the Successive Projection Algorithm
• PD-ML-Lite: Private Distributed Machine Learning from Lighweight Cryptography
• Endmember Extraction from Hyperspectral Images Using Self-Dictionary Approach with Linear Programming
• Deep topic modeling by multilayer bootstrap network and lasso