Robust Volume Minimization-Based Matrix Factorization for Remote Sensing and Document Clustering

Abstract

This paper considers \emph{volume minimization} (VolMin)-based structured matrix factorization (SMF). VolMin is a factorization criterion that decomposes a given data matrix into a basis matrix times a structured coefficient matrix via finding the minimum-volume simplex that encloses all the columns of the data matrix. Recent work showed that VolMin guarantees the identifiability of the factor matrices under mild conditions that are realistic in a wide variety of applications. This paper focuses on both theoretical and practical aspects of VolMin. On the theory side, exact equivalence of two independently developed sufficient conditions for VolMin identifiability is proven here, thereby providing a more comprehensive understanding of this aspect of VolMin. On the algorithm side, computational complexity and sensitivity to outliers are two key challenges associated with real-world applications of VolMin. These are addressed here via a new VolMin algorithm that handles volume regularization in a computationally simple way, and automatically detects and {iteratively downweights} outliers, simultaneously. Simulations and real-data experiments using a remotely sensed hyperspectral image and the Reuters document corpus are employed to showcase the effectiveness of the proposed algorithm.

References in corpus (10)

• Robust Recovery of Subspace Structures by Low-Rank Representation
• A Variable Splitting Augmented Lagrangian Approach to Linear Spectral Unmixing
• Nonlinear unmixing of hyperspectral images: models and algorithms
• Fast and Robust Recursive Algorithms for Separable Nonnegative Matrix Factorization
• Nonlinear hyperspectral unmixing with robust nonnegative matrix factorization
• Projection onto the probability simplex: An efficient algorithm with a simple proof, and an application
• Identifiability of the Simplex Volume Minimization Criterion for Blind Hyperspectral Unmixing: The No Pure-Pixel Case
• Self-Dictionary Sparse Regression for Hyperspectral Unmixing: Greedy Pursuit and Pure Pixel Search are Related
• Joint Tensor Factorization and Outlying Slab Suppression with Applications
• A Smoothing SQP Framework for a Class of Composite $L_q$ Minimization over Polyhedron

Cited by in corpus (32)

• Hyperspectral Super-Resolution: A Coupled Tensor Factorization Approach
• Nonnegative Matrix Factorization for Signal and Data Analytics: Identifiability, Algorithms, and Applications
• Deep Hyperspectral Unmixing using Transformer Network
• On Identifiability of Nonnegative Matrix Factorization
• Blind Audio Source Separation with Minimum-Volume Beta-Divergence NMF
• Fast Low-Rank Bayesian Matrix Completion with Hierarchical Gaussian Prior Models
• Learning From Hidden Traits: Joint Factor Analysis and Latent Clustering
• Algorithms and Comparisons of Non-negative Matrix Factorization with Volume Regularization for Hyperspectral Unmixing
• Spectral Variability Aware Blind Hyperspectral Image Unmixing Based on Convex Geometry
• Accelerating Nonnegative Matrix Factorization Algorithms using Extrapolation
• Penalty Dual Decomposition Method For Nonsmooth Nonconvex Optimization
• Generalized Separable Nonnegative Matrix Factorization
• Computing Large-Scale Matrix and Tensor Decomposition with Structured Factors: A Unified Nonconvex Optimization Perspective
• A consistent and flexible framework for deep matrix factorizations
• Provably End-to-end Label-Noise Learning without Anchor Points
• Inexact Alternating Optimization for Phase Retrieval In the Presence of Outliers
• Simplex-Structured Matrix Factorization: Sparsity-based Identifiability and Provably Correct Algorithms
• Nonlinear Multiview Analysis: Identifiability and Neural Network-assisted Implementation
• Recovering Joint Probability of Discrete Random Variables from Pairwise Marginals
• Hybrid Inexact BCD for Coupled Structured Matrix Factorization in Hyperspectral Super-Resolution
• Probabilistic Simplex Component Analysis
• Polytopic Matrix Factorization: Determinant Maximization Based Criterion and Identifiability
• Memory-Efficient Convex Optimization for Self-Dictionary Separable Nonnegative Matrix Factorization: A Frank-Wolfe Approach
• Learning Nonlinear Mixtures: Identifiability and Algorithm
• Near-Convex Archetypal Analysis
• Maximum Volume Inscribed Ellipsoid: A New Simplex-Structured Matrix Factorization Framework via Facet Enumeration and Convex Optimization
• Identifiability-Guaranteed Simplex-Structured Post-Nonlinear Mixture Learning via Autoencoder
• Mixed Membership Graph Clustering via Systematic Edge Query
• Checking the Sufficiently Scattered Condition using a Global Non-Convex Optimization Software
• Extracting Optimal Solution Manifolds using Constrained Neural Optimization
• Communication-Efficient Federated Linear and Deep Generalized Canonical Correlation Analysis
• Crowdsourcing via Annotator Co-occurrence Imputation and Provable Symmetric Nonnegative Matrix Factorization