The multivariate Watson distribution: maximum-likelihood estimation and other aspects

DOI10.1016/J.JMVA.2012.08.010MaRDI QIDQ1931867zbMATH OpenOpenAlexFDO

Authors Suvrit Sra, Dmitry B. Karp

Publication date 16 January 2013

Published in Journal of Multivariate Analysis (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1104.4422

zbMATH Keywords

directional statisticsconfluent hypergeometricspecial functionhypergeometric identitiesWatson distributionKummer functiondiametrical clustering

Mathematics Subject Classification ID

Directional data; spatial statistics (62H11) Classification and discrimination; cluster analysis (statistical aspects) (62H30) Estimation in multivariate analysis (62H12) Confluent hypergeometric functions, Whittaker functions, ({}_1F_1) (33C15) Applications of hypergeometric functions (33C90)

Abstract: This paper studies fundamental aspects of modelling data using multivariate Watson distributions. Although these distributions are natural for modelling axially symmetric data (i.e., unit vectors where $pmx$ are equivalent), for high-dimensions using them can be difficult. Why so? Largely because for Watson distributions even basic tasks such as maximum-likelihood are numerically challenging. To tackle the numerical difficulties some approximations have been derived---but these are either grossly inaccurate in high-dimensions (emph{Directional Statistics}, Mardia & Jupp. 2000) or when reasonably accurate (emph{J. Machine Learning Research, W. & C.P., v2}, Bijral emph{et al.}, 2007, pp. 35--42), they lack theoretical justification. We derive new approximations to the maximum-likelihood estimates; our approximations are theoretically well-defined, numerically accurate, and easy to compute. We build on our parameter estimation and discuss mixture-modelling with Watson distributions; here we uncover a hitherto unknown connection to the "diametrical clustering" algorithm of Dhillon emph{et al.} (emph{Bioinformatics}, 19(13), 2003, pp. 1612--1619).

Recommendations

• Inference on parameters of Watson distributions and application to classification of observations
• Multi-sample tests for axial data from Watson distributions
• Discriminant analysis based on the Watson distribution defined on the hypersphere
• Test for Rotational Symmetry of Data from the Watson Distribution Defined on the Hypersphere
• Goodness-of-fit methods for the bipolar Watson distribution defined on the hypersphere

Cited in

(19)

• Distance-based hypothesis tests on the Watson distribution
• Generalised Watson distribution on the hypersphere with applications to clustering
• Clustering of variables based on Watson distribution on hypersphere: a comparison of algorithms
• Information-theoretic registration with explicit reorientation of diffusion-weighted images
• Monotonicity patterns and functional inequalities for the Nuttall \(Q\)-function
• Recent advances in directional statistics
• scientific article; zbMATH DE number 7370518 (Why is no real title available?)
• On the MLE of the Waring distribution
• Inference on parameters of Watson distributions and application to classification of observations
• Poisson Kernel-Based Clustering on the Sphere: Convergence Properties, Identifiability, and a Method of Sampling
• Testing uniformity on high-dimensional spheres: the non-null behaviour of the Bingham test
• Log-convexity and log-concavity for series in gamma ratios and applications
• On the power of axial tests of uniformity on spheres
• Family of integrable bounds for the logarithmic derivative of Kummer's function
• Log-concavity for series in reciprocal gamma functions and applications
• Universal approximation on the hypersphere
• Model-based clustering of probability density functions
• Sharp bounds for cumulative distribution functions
• On bounds for Kummer's function ratio

This page was built for publication: The multivariate Watson distribution: maximum-likelihood estimation and other aspects

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1931867)