Image labeling based on graphical models using Wasserstein messages and geometric assignment
From MaRDI portal
Abstract: We introduce a novel approach to Maximum A Posteriori inference based on discrete graphical models. By utilizing local Wasserstein distances for coupling assignment measures across edges of the underlying graph, a given discrete objective function is smoothly approximated and restricted to the assignment manifold. A corresponding multiplicative update scheme combines in a single process (i) geometric integration of the resulting Riemannian gradient flow and (ii) rounding to integral solutions that represent valid labelings. Throughout this process, local marginalization constraints known from the established LP relaxation are satisfied, whereas the smooth geometric setting results in rapidly converging iterations that can be carried out in parallel for every edge.
Recommendations
- MAP image labeling using Wasserstein messages and geometric assignment
- Image labeling by assignment
- Continuous multiclass labeling approaches and algorithms
- Learning adaptive regularization for image labeling using geometric assignment
- Graphical model parameter learning by inverse linear programming
Cites work
- A D.C. Optimization Algorithm for Solving the Trust-Region Subproblem
- A Graph Framework for Manifold-Valued Data
- A Linear Programming Formulation and Approximation Algorithms for the Metric Labeling Problem
- A Relationship Between Arbitrary Positive Matrices and Doubly Stochastic Matrices
- A smoothed dual approach for variational Wasserstein problems
- A sparse multiscale algorithm for dense optimal transport
- Approximate classification via earthmover metrics
- Approximation algorithms for classification problems with pairwise relationships, metric labeling and Markov random fields
- Combinatorial matrix classes
- Constructing Free-Energy Approximations and Generalized Belief Propagation Algorithms
- Convex analysis approach to d. c. programming: Theory, algorithms and applications
- Diffuse Interface Models on Graphs for Classification of High Dimensional Data
- Entropic approximation of Wasserstein gradient flows
- Estimation and Marginalization Using the Kikuchi Approximation Methods
- Geometry of cuts and metrics
- Graphical models, exponential families, and variational inference
- scientific article; zbMATH DE number 3790208 (Why is no real title available?)
- scientific article; zbMATH DE number 729680 (Why is no real title available?)
- scientific article; zbMATH DE number 1046019 (Why is no real title available?)
- scientific article; zbMATH DE number 2062604 (Why is no real title available?)
- scientific article; zbMATH DE number 1849127 (Why is no real title available?)
- scientific article; zbMATH DE number 2107836 (Why is no real title available?)
- Image labeling by assignment
- Interior point methods of mathematical programming
- Iterative multiplicative filters for data labeling
- Linear programming relaxations and belief propagation -- an empirical study
- Linear programming, complexity theory and elementary functional analysis
- MAP Estimation Via Agreement on Trees: Message-Passing and Linear Programming
- MAP image labeling using Wasserstein messages and geometric assignment
- Matrix scaling, entropy minimization, and conjugate duality. II: The dual problem
- Message-passing for graph-structured linear programs: proximal methods and rounding schemes
- Monotone Operators and the Proximal Point Algorithm
- Nonlocal discrete p-Laplacian driven image and manifold processing
- Nonlocal Operators with Applications to Image Processing
- Norm-Product Belief Propagation: Primal-Dual Message-Passing for Approximate Inference
- Numerical integration of Riemannian gradient flows for image labeling
- On a special class of convex functions
- Proximal minimization algorithm with \(D\)-functions
- Riemannian center of mass and mollifier smoothing
- The Boolean quadratic polytope: Some characteristics, facets and relatives
- The Sinkhorn–Knopp Algorithm: Convergence and Applications
- The Theory of Max-Min, with Applications
- Variational Analysis
- Variational problems in SBV and image segmentation
Cited in
(8)- Image labeling by assignment
- Learning adaptive regularization for image labeling using geometric assignment
- Assignment flows
- Geometric latent Dirichlet allocation on a matching graph for large-scale image datasets
- Geometric numerical integration of the assignment flow
- Graphical model parameter learning by inverse linear programming
- MAP image labeling using Wasserstein messages and geometric assignment
- Proxying credit curves via Wasserstein distances
This page was built for publication: Image labeling based on graphical models using Wasserstein messages and geometric assignment
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4689763)
