English

Bayesian Hyperbolic Multidimensional Scaling

Methodology 2023-08-16 v3 Machine Learning Computation Machine Learning

Abstract

Multidimensional scaling (MDS) is a widely used approach to representing high-dimensional, dependent data. MDS works by assigning each observation a location on a low-dimensional geometric manifold, with distance on the manifold representing similarity. We propose a Bayesian approach to multidimensional scaling when the low-dimensional manifold is hyperbolic. Using hyperbolic space facilitates representing tree-like structures common in many settings (e.g. text or genetic data with hierarchical structure). A Bayesian approach provides regularization that minimizes the impact of measurement error in the observed data and assesses uncertainty. We also propose a case-control likelihood approximation that allows for efficient sampling from the posterior distribution in larger data settings, reducing computational complexity from approximately O(n2)O(n^2) to O(n)O(n). We evaluate the proposed method against state-of-the-art alternatives using simulations, canonical reference datasets, Indian village network data, and human gene expression data.

Keywords

Cite

@article{arxiv.2210.15081,
  title  = {Bayesian Hyperbolic Multidimensional Scaling},
  author = {Bolun Liu and Shane Lubold and Adrian E. Raftery and Tyler H. McCormick},
  journal= {arXiv preprint arXiv:2210.15081},
  year   = {2023}
}