Geometric numerical integration of the assignment flow
From MaRDI portal
(Redirected from Publication:5000560)
adaptive step size selectionassignment flowassignment manifoldgeometric integrationimage labelingKrylov subspace approximation
Image analysis in multivariate analysis (62H35) Iterative numerical methods for linear systems (65F10) Numerical methods for initial value problems involving ordinary differential equations (65L05) Multistep, Runge-Kutta and extrapolation methods for ordinary differential equations (65L06) Learning and adaptive systems in artificial intelligence (68T05) Image processing (compression, reconstruction, etc.) in information and communication theory (94A08)
Abstract: The assignment flow is a smooth dynamical system that evolves on an elementary statistical manifold and performs contextual data labeling on a graph. We derive and introduce the linear assignment flow that evolves nonlinearly on the manifold, but is governed by a linear ODE on the tangent space. Various numerical schemes adapted to the mathematical structure of these two models are designed and studied, for the geometric numerical integration of both flows: embedded Runge-Kutta-Munthe-Kaas schemes for the nonlinear flow, adaptive Runge-Kutta schemes and exponential integrators for the linear flow. All algorithms are parameter free, except for setting a tolerance value that specifies adaptive step size selection by monitoring the local integration error, or fixing the dimension of the Krylov subspace approximation. These algorithms provide a basis for applying the assignment flow to machine learning scenarios beyond supervised labeling, including unsupervised labeling and learning from controlled assignment flows.
Recommendations
- Geometric integrators for ODEs
- Geometric integration algorithms on homogeneous manifolds
- scientific article; zbMATH DE number 4213749
- scientific article; zbMATH DE number 1745051
- Geometric Numerical Integration
- Geometric Numerical Integration
- scientific article; zbMATH DE number 1343130
- Geometric integration: numerical solution of differential equations on manifolds
- scientific article; zbMATH DE number 1734659
- Optimal geometric flows via dual programs
Cites work
- A proposal on machine learning via dynamical systems
- Algorithm 919: A Krylov subspace algorithm for evaluating the -functions appearing in exponential integrators
- Analysis of Some Krylov Subspace Approximations to the Matrix Exponential Operator
- Computing the action of the matrix exponential, with an application to exponential integrators
- Expokit
- Exponential integrators
- Functions of Matrices
- Geometric Numerical Integration
- High order Runge-Kutta methods on manifolds
- scientific article; zbMATH DE number 1560711 (Why is no real title available?)
- scientific article; zbMATH DE number 1795895 (Why is no real title available?)
- Image labeling based on graphical models using Wasserstein messages and geometric assignment
- Image labeling by assignment
- Information geometry
- Iterative multiplicative filters for data labeling
- Nineteen Dubious Ways to Compute the Exponential of a Matrix, Twenty-Five Years Later
- On a notion of linear replicator equations
- On Krylov Subspace Approximations to the Matrix Exponential Operator
- On some inequalities for the incomplete gamma function
- Ordinary differential equations and dynamical systems
- Solving Ordinary Differential Equations I
- Stable architectures for deep neural networks
Cited in
(18)- Learning adaptive regularization for image labeling using geometric assignment
- Assignment flow for order-constrained OCT segmentation
- Assignment flows for data labeling on graphs: convergence and stability
- Assignment flows
- scientific article; zbMATH DE number 4213749 (Why is no real title available?)
- Continuous-domain assignment flows
- Self-assignment flows for unsupervised data labeling on graphs
- Numerical integration of Riemannian gradient flows for image labeling
- On Multiscale Quasi-Interpolation of Scattered Scalar- and Manifold-Valued Functions
- Learning linearized assignment flows for image labeling
- A Nonlocal Graph-PDE and Higher-Order Geometric Integration for Image Labeling
- A variational perspective on the assignment flow
- On the geometric mechanics of assignment flows for metric data labeling
- Riemannian patch assignment gradient flows
- Generative assignment flows for representing and learning joint distributions of discrete data
- On the correspondence between replicator dynamics and assignment flows
- Learning linear assignment flows for image labeling via exponential integration
- On the geometric mechanics of assignment flows for metric data labeling
This page was built for publication: Geometric numerical integration of the assignment flow
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5000560)
