Abstract:Data from individual observations can originate from various sources or modalities but are often intrinsically linked. Multimodal data integration can enrich information content compared to single-source data. Manifold alignment is a form of data integration that seeks a shared, underlying low-dimensional representation of multiple data sources that emphasizes similarities between alternative representations of the same entities. Semi-supervised manifold alignment relies on partially known correspondences between domains, either through shared features or through other known associations. In this paper, we introduce two semi-supervised manifold alignment methods. The first method, Shortest Paths on the Union of Domains (SPUD), forms a unified graph structure using known correspondences to establish graph edges. By learning inter-domain geodesic distances, SPUD creates a global, multi-domain structure. The second method, MASH (Manifold Alignment via Stochastic Hopping), learns local geometry within each domain and forms a joint diffusion operator using known correspondences to iteratively learn new inter-domain correspondences through a random-walk approach. Through the diffusion process, MASH forms a coupling matrix that links heterogeneous domains into a unified structure. We compare SPUD and MASH with existing semi-supervised manifold alignment methods and show that they outperform competing methods in aligning true correspondences and cross-domain classification. In addition, we show how these methods can be applied to transfer label information between domains.

Semi-definite Manifold Alignment

Learning Visually Aligned Semantic Graph for Cross-Modal Manifold Matching.

Manifold Alignment Via Corresponding Projections.

Learning Exemplar-Represented Manifolds in Latent Space for Classification.

Manifold Alignment Via Global and Local Structures Preserving PCA Framework.

Graph Integration for Diffusion-Based Manifold Alignment

Unsupervised image matching based on manifold alignment.

Deep Manifold Computing and Visualization

Spectral Methods for Semi-Supervised Manifold Learning.

A General Framework for 3D Model Co-Alignment.

Geodesic Based Semi-supervised Multi-manifold Feature Extraction

Random Forest-Supervised Manifold Alignment

ManiDec: Manifold Constrained Low-Rank and Sparse Decomposition

Submanifold Decomposition

A Regularized Approach for Geodesic-Based Semisupervised Multimanifold Learning

Scalable unsupervised alignment of general metric and non-metric structures

Supervised Semi-Definite Embedding For Image Manifolds

Ensemble Manifold Regularization

Manifold-manifold distance and its application to face recognition with image sets.

Local Orthogonality Preserving Alignment for Nonlinear Dimensionality Reduction

Local Coordinates Alignment (Lca): A Novel Manifold Learning Approach