Horseshoes in multidimensional scaling and local kernel methods
From MaRDI portal
Publication:958313
Abstract: Classical multidimensional scaling (MDS) is a method for visualizing high-dimensional point clouds by mapping to low-dimensional Euclidean space. This mapping is defined in terms of eigenfunctions of a matrix of interpoint dissimilarities. In this paper we analyze in detail multidimensional scaling applied to a specific dataset: the 2005 United States House of Representatives roll call votes. Certain MDS and kernel projections output ``horseshoes that are characteristic of dimensionality reduction techniques. We show that, in general, a latent ordering of the data gives rise to these patterns when one only has local information. That is, when only the interpoint distances for nearby points are known accurately. Our results provide a rigorous set of results and insight into manifold learning in the special case where the manifold is a curve.
Recommendations
- Multidimensional scaling of noisy high dimensional data
- Subspace Least Squares Multidimensional Scaling
- scientific article; zbMATH DE number 912679
- Stress functions for nonlinear dimension reduction, proximity analysis, and graph drawing
- Taking all positive eigenvectors is suboptimal in classical multidimensional scaling
Cites work
- A general nonmetric technique for finding the smallest coordinate space for a configuration of points
- Centrosymmetric (Cross-Symmetric) Matrices, Their Basic Properties, Eigenvalues, and Eigenvectors
- Consistency of spectral clustering
- Eigenvalues and eigenvectors of symmetric centrosymmetric matrices
- Horseshoes in multidimensional scaling and local kernel methods
- scientific article; zbMATH DE number 5286897 (Why is no real title available?)
- scientific article; zbMATH DE number 3903772 (Why is no real title available?)
- scientific article; zbMATH DE number 3671573 (Why is no real title available?)
- scientific article; zbMATH DE number 3673370 (Why is no real title available?)
- Laplacian Eigenmaps for Dimensionality Reduction and Data Representation
- Modern multidimensional scaling: theory and applications
- Multidimensional scaling. I: Theory and method
- Remarks to Maurice Frechet's article ``Sur la definition axiomatique d'une classe d'espaces vectoriels distancies applicables vectoriellement sur l'espace de Hilbert
- Spectral properties of distance matrices
- The analysis of proximities: Multidimensional scaling with an unknown distance function: I, II
- The Inverse of a Centrosymmetric Matrix
Cited in
(19)- Data spectroscopy: eigenspaces of convolution operators and clustering
- Spectral properties of Kac-Murdock-Szegö matrices with a complex parameter
- On Rayleigh-type formulas for a non-local boundary value problem associated with an integral operator commuting with the Laplacian
- Multidimensional scaling on metric measure spaces
- Exact recovery in the Ising blockmodel
- Distributions of eigenvalues of large Euclidean matrices generated from l_p balls and spheres
- Multiple correspondence analysis and the multilogit bilinear model
- Separable factor analysis with applications to mortality data
- Spectrum of large Euclidean random matrices generated from \(l_p\) ellipsoids
- Statistical learning on emerging economies
- Interpreting uninterpretable predictors: kernel methods, Shtarkov solutions, and random forests
- Theoretical analysis of principal components in an umbrella model of intraspecific evolution
- Non-parametric regression for networks
- On the approximation of kernel functions
- Correspondence Analysis on Sparse Bipartite Graphs with Hyperspecialization
- Euclidean mirrors and first-order changepoints in network time series
- Estimating heterogeneous graphical models for discrete data with an application to roll call voting
- Graphics processing units and high-dimensional optimization
- Horseshoes in multidimensional scaling and local kernel methods
This page was built for publication: Horseshoes in multidimensional scaling and local kernel methods
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q958313)
