Low-Rank Dependence Decomposition via Accelerated Symmetric Non-negative Matrix Factorization

DOI10.48550/arXiv.2607.24518MaRDI QIDQ7284410arXivFDO

Authors Lavinia Ghita, Dhruv Desai, Jake Goldberg, Roman Yokunda Enzmann

Publication date 27 July 2026

Published in arXiv

Tldr We reformulate SymNMF and test seven algorithms on extreme-scale dependence matrices; we accelerate SymNMF to scale beyond 10^6 variables with five efficient methods that converge at very large sizes.

Full work available at URL https://doi.org/10.48550/arXiv.2607.24518

arXiv classification

cs.LGmath.NA

Cites work

• A dependence measure for multivariate and spatial extreme values: Properties and inference
• A Fast Iterative Shrinkage-Thresholding Algorithm for Linear Inverse Problems
• A method for the solution of certain non-linear problems in least squares
• A well-conditioned estimator for large-dimensional covariance matrices
• Acceleration of Stochastic Approximation by Averaging
• Adan: Adaptive Nesterov Momentum Algorithm for Faster Optimizing Deep Models
• Adaptive restart for accelerated gradient schemes
• Adaptive subgradient methods for online learning and stochastic optimization
• Algorithms for nonnegative matrix and tensor factorizations: a unified view based on block coordinate descent framework
• An Algorithm for Least-Squares Estimation of Nonlinear Parameters
• An introduction to statistical modeling of extreme values
• Bivariate extreme statistics. I
• Concept decompositions for large sparse text data using clustering
• Consistency of spectral clustering in stochastic block models
• Decompositions of dependence for high-dimensional extremes
• Distributed optimization and statistical learning via the alternating direction method of multipliers
• DISTRIBUTION OF EIGENVALUES FOR SOME SETS OF RANDOM MATRICES
• Dual averaging methods for regularized stochastic learning and online optimization
• Extremal behavior of stochastic integrals driven by regularly varying Lévy processes
• Extreme Value Theory
• Finding structure with randomness: probabilistic algorithms for constructing approximate matrix decompositions
• Grad-GradaGrad? A Non-Monotone Adaptive Stochastic Gradient Method
• Hybrid deterministic-stochastic methods for data fitting
• Learning the parts of objects by non-negative matrix factorization
• Minimization of functions having Lipschitz continuous first partial derivatives
• Multivariate regularly varying insurance and financial risks in multidimensional risk models
• Nonlinear Programming
• Nonnegative matrix factorization
• Nonnegative Matrix Factorization Based on Alternating Nonnegativity Constrained Least Squares and Active Set Method
• Numerical Optimization
• Projected Gradient Methods for Nonnegative Matrix Factorization
• Randomized algorithms for symmetric nonnegative matrix factorization
• Sample size selection in optimization methods for machine learning
• SGDR: Stochastic Gradient Descent with Warm Restarts
• Shampoo: Preconditioned Stochastic Tensor Optimization
• Sketching as a tool for numerical linear algebra
• Some methods of speeding up the convergence of iteration methods
• Spherical k-Means++ Clustering
• SVD based initialization: A head start for nonnegative matrix factorization
• SymNMF: nonnegative low-rank approximation of a similarity matrix for graph clustering

This page was built for publication: Low-Rank Dependence Decomposition via Accelerated Symmetric Non-negative Matrix Factorization

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7284410)