Wassmap: Wasserstein Isometric Mapping for Image Manifold Learning
From MaRDI portal
Abstract: In this paper, we propose Wasserstein Isometric Mapping (Wassmap), a nonlinear dimensionality reduction technique that provides solutions to some drawbacks in existing global nonlinear dimensionality reduction algorithms in imaging applications. Wassmap represents images via probability measures in Wasserstein space, then uses pairwise Wasserstein distances between the associated measures to produce a low-dimensional, approximately isometric embedding. We show that the algorithm is able to exactly recover parameters of some image manifolds including those generated by translations or dilations of a fixed generating measure. Additionally, we show that a discrete version of the algorithm retrieves parameters from manifolds generated from discrete measures by providing a theoretical bridge to transfer recovery results from functional data to discrete data. Testing of the proposed algorithms on various image data manifolds show that Wassmap yields good embeddings compared with other global and local techniques.
Recommendations
- Linearized Wasserstein dimensionality reduction with approximation guarantees
- Continuum isomap for manifold learnings
- PARAMAP vs. Isomap: a comparison of two nonlinear mapping algorithms
- Nonlinear dimensionality reduction by topologically constrained isometric embedding
- Image manifolds which are isometric to Euclidean space
Cites work
- A linear optimal transportation framework for quantifying and visualizing variations in sets of images
- A sparse multiscale algorithm for dense optimal transport
- Computational optimal transport. With applications to data sciences
- Diffusion maps
- Domain decomposition for entropy regularized optimal transport
- Fast low-rank modifications of the thin singular value decomposition
- Hessian eigenmaps: Locally linear embedding techniques for high-dimensional data
- scientific article; zbMATH DE number 3673370 (Why is no real title available?)
- scientific article; zbMATH DE number 1909499 (Why is no real title available?)
- Image manifolds which are isometric to Euclidean space
- Iterative scheme for solving optimal transportation problems arising in reflector design
- Laplacian Eigenmaps for Dimensionality Reduction and Data Representation
- Linear optimal transport embedding: provable Wasserstein classification for certain rigid transformations and perturbations
- Manifold learning with arbitrary norms
- Manifold parametrizations by eigenfunctions of the Laplacian and heat kernels
- Multiscale geometric methods for data sets. I: Multiscale SVD, noise and curvature.
- Multiscale strategies for computing optimal transport
- Nonlinear Dimensionality Reduction
- Nonlinear manifold learning for early warnings in financial markets
- Nonlinear manifold representations for functional data
- On the geometry of metric measure spaces. I
- Optimal Transport
- Optimal transport for applied mathematicians. Calculus of variations, PDEs, and modeling
- Partitioning signal classes using transport transforms for data analysis and machine learning
- Polar factorization and monotone rearrangement of vector‐valued functions
- Reducing the Dimensionality of Data with Neural Networks
- Revisiting the Nyström method for improved large-scale machine learning
- Stabilized Sparse Scaling Algorithms for Entropy Regularized Transport Problems
- Supervised learning of sheared distributions using linearized optimal transport
- Testing the manifold hypothesis
- The cumulative distribution transform and linear pattern classification
- The geometry of optimal transportation
- The Radon Cumulative Distribution Transform and Its Application to Image Classification
- Visualizing data using t-SNE
- Wasserstein proximal of GANs
Cited in
(10)- Applications of No-Collision Transportation Maps in Manifold Learning
- On Wasserstein distances for affine transformations of random vectors
- Linearized Wasserstein dimensionality reduction with approximation guarantees
- Manifold learning in metric spaces
- Joint manifold learning and optimal transport for dynamic imaging
- Visualizing shape functionals via Sinkhorn multidimensional scaling
- Manifold learning in Wasserstein space
- Embedding distributional data
- Linear independent component analysis in Wasserstein space
- Supervised dimension reduction via local gradient elongation
This page was built for publication: Wassmap: Wasserstein Isometric Mapping for Image Manifold Learning
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6171687)
