Wasserstein dictionary learning: optimal transport-based unsupervised nonlinear dictionary learning
From MaRDI portal
Abstract: This paper introduces a new nonlinear dictionary learning method for histograms in the probability simplex. The method leverages optimal transport theory, in the sense that our aim is to reconstruct histograms using so-called displacement interpolations (a.k.a. Wasserstein barycenters) between dictionary atoms; such atoms are themselves synthetic histograms in the probability simplex. Our method simultaneously estimates such atoms, and, for each datapoint, the vector of weights that can optimally reconstruct it as an optimal transport barycenter of such atoms. Our method is computationally tractable thanks to the addition of an entropic regularization to the usual optimal transportation problem, leading to an approximation scheme that is efficient, parallel and simple to differentiate. Both atoms and weights are learned using a gradient-based descent method. Gradients are obtained by automatic differentiation of the generalized Sinkhorn iterations that yield barycenters with entropic smoothing. Because of its formulation relying on Wasserstein barycenters instead of the usual matrix product between dictionary and codes, our method allows for nonlinear relationships between atoms and the reconstruction of input data. We illustrate its application in several different image processing settings.
Recommendations
Cites work
- scientific article; zbMATH DE number 3829300 (Why is no real title available?)
- scientific article; zbMATH DE number 107616 (Why is no real title available?)
- scientific article; zbMATH DE number 1465030 (Why is no real title available?)
- scientific article; zbMATH DE number 1909499 (Why is no real title available?)
- scientific article; zbMATH DE number 3231692 (Why is no real title available?)
- scientific article; zbMATH DE number 3053873 (Why is no real title available?)
- rm K-SVD: An Algorithm for Designing Overcomplete Dictionaries for Sparse Representation
- A Direct Formulation for Sparse PCA Using Semidefinite Programming
- A class of Cartesian grid embedded boundary algorithms for incompressible flow with time-varying complex geometries
- A convexity principle for interacting gases
- A linear optimal transportation framework for quantifying and visualizing variations in sets of images
- A smoothed dual approach for variational Wasserstein problems
- A survey of the Schrödinger problem and some of its connections with optimal transport
- Asymptotic Properties and Robustness of Minimum Dissimilarity Estimators of Location-scale Parameters
- Barycenters in the Wasserstein space
- Convolutional Wasserstein distances: efficient optimal transportation on geometric domains
- Design of Non-Linear Kernel Dictionaries for Object Recognition
- Diagonal Equivalence to Matrices with Prescribed Row and Column Sums
- Distribution's template estimate with Wasserstein metrics
- Evaluating Derivatives
- Fast Discrete Distribution Clustering Using Wasserstein Barycenter With Sparse Support
- Geodesic PCA in the Wasserstein space by convex PCA
- Heavy-ball method in nonconvex optimization problems
- Introductory lectures on convex optimization. A basic course.
- Iterative Bregman projections for regularized transportation problems
- Learning the parts of objects by non-negative matrix factorization
- Mass transportation problems. Vol. 1: Theory. Vol. 2: Applications
- Numerical methods for matching for teams and Wasserstein barycenters
- On a Least Squares Adjustment of a Sampled Frequency Table When the Expected Marginal Totals are Known
- On minimum Kantorovich distance estimators
- On the scaling of multidimensional matrices
- Online Dictionary Learning for Kernel LMS
- Online learning for matrix factorization and sparse coding
- Optimal mass transport for registration and warping
- Point spread function field learning based on optimal transport distances
- Reducing the Dimensionality of Data with Neural Networks
- Remark on ``Algorithm 778: L-BFGS-B: Fortran subroutines for large-scale bound constrained optimization
- Sliced and Radon Wasserstein barycenters of measures
- Some methods of speeding up the convergence of iteration methods
- The auction algorithm: A distributed relaxation method for the assignment problem
- The earth mover's distance as a metric for image retrieval
- Transportation cost for Gaussian and other product measures
- Wasserstein dictionary learning: optimal transport-based unsupervised nonlinear dictionary learning
Cited in
(30)- Wasserstein model reduction approach for parametrized flow problems in porous media
- Randomized Wasserstein barycenter computation: resampling with statistical guarantees
- A column generation approach to the discrete barycenter problem
- Isometries of the qubit state space with respect to quantum Wasserstein distances
- Wasserstein dictionary learning: optimal transport-based unsupervised nonlinear dictionary learning
- Unbalanced Kantorovich-Rubinstein distance, plan, and barycenter on finite spaces: a statistical perspective
- Shared Prior Learning of Energy-Based Models for Image Reconstruction
- Augmented projection Wasserstein distances: multi-dimensional projection with neural surface
- Learning to generate Wasserstein barycenters
- Projection-based techniques for high-dimensional optimal transport problems
- Ground metric learning on graphs
- Learning finite-dimensional coding schemes with nonlinear reconstruction maps
- A registration method for reduced basis problems using linear optimal transport
- Empirical regularized optimal transport: statistical theory and applications
- Minimax estimation of smooth optimal transport maps
- Asymptotics for Strassen's optimal transport problem
- Semi-discrete optimal transport: a solution procedure for the unsquared Euclidean distance case
- Metrics and barycenters for point pattern data
- Central limit theorems for entropy-regularized optimal transport on finite spaces and statistical applications
- A general drift estimation procedure for stochastic differential equations with additive fractional noise
- scientific article; zbMATH DE number 7370538 (Why is no real title available?)
- Rethinking data-driven point spread function modeling with a differentiable optical model
- Estimation of Wasserstein distances in the spiked transport model
- Wasserstein index generation model: automatic generation of time-series index with application to economic policy uncertainty
- A family of pairwise multi-marginal optimal transports that define a generalized metric
- Regularized optimal transport and the rot mover's distance
- Multigrid methods for image registration model based on optimal mass transport
- Joint manifold learning and optimal transport for dynamic imaging
- Sparse Wasserstein barycenters and application to reduced order modeling
- Kantorovich-Rubinstein distance and barycenter for finitely supported measures: foundations and algorithms
This page was built for publication: Wasserstein dictionary learning: optimal transport-based unsupervised nonlinear dictionary learning
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4686929)
