Tukey depths and Hamilton-Jacobi differential equations

DOI10.1137/21M1411998zbMATH Open1490.35478arXiv2104.01648OpenAlexW3144903544WikidataQ115246859 ScholiaQ115246859MaRDI QIDQ5075722FDOQ5075722

Authors: M. Molina-Fructuoso, Ryan W. Murray 

Publication date: 11 May 2022

Published in: SIAM Journal on Mathematics of Data Science (Search for Journal in Brave)

Abstract: The widespread application of modern machine learning has increased the need for robust statistical algorithms. This work studies one such fundamental statistical measure known as the Tukey depth. We study the problem in the continuum (population) limit. In particular, we derive the associated necessary conditions, which take the form of a first-order partial differential equation. We discuss the classical interpretation of this necessary condition as the viscosity solution of a Hamilton-Jacobi equation, but with a non-classical Hamiltonian with discontinuous dependence on the gradient at zero. We prove that this equation possesses a unique viscosity solution and that this solution always bounds the Tukey depth from below. In certain cases, we prove that the Tukey depth is equal to the viscosity solution, and we give some illustrations of standard numerical methods from the optimal control community which deal directly with the partial differential equation. We conclude by outlining several promising research directions both in terms of new numerical algorithms and theoretical challenges.

Full work available at URL: https://arxiv.org/abs/2104.01648

Recommendations

• Fast implementation of the Tukey depth
• General notions of statistical depth function.
• Concentration of the empirical level sets of Tukey's halfspace depth
• On depth and deep points: A calculus.
• Some extensions of Tukey's depth function

zbMATH Keywords

robust statisticsHamilton-Jacobi equationsviscosity solutionsstatistical depths

Mathematics Subject Classification ID

Nonparametric robustness (62G35) Characterization and structure theory for multivariate probability distributions; copulas (62H05) Combinatorial probability (60C05) Hamilton-Jacobi equations (35F21) PDEs in connection with statistics (35Q62)

Cites Work

• Breakdown properties of location estimates based on halfspace depth and projected outlyingness
• Robust estimation and classification for functional data via projection-based depth notions
• The random Tukey depth
• General notions of statistical depth function.
• Clustering and classification based on the L\(_{1}\) data depth
• Projection-based depth functions and associated medians
• Exact computation of the halfspace depth
• Title not available (Why is that?)
• Propagation of singularities for solutions of nonlinear first order partial differential equations
• Title not available (Why is that?)
• Viscosity Solutions of Hamilton-Jacobi Equations
• User’s guide to viscosity solutions of second order partial differential equations
• Computing multiple-output regression quantile regions
• A Bivariate Sign Test
• The depth function of a population distribution.
• Smooth depth contours characterize the underlying distribution
• A Simple, Direct Proof of Uniqueness for Solutions of the Hamilton-Jacobi Equations of Eikonal Type
• Title not available (Why is that?)
• Nonparametric multivariate kurtosis and tailweight measures
• Eikonal equation adaptation on weighted graphs: fast geometric diffusion process for local and non-local image and data processing
• Fast methods for the Eikonal and related Hamilton– Jacobi equations on unstructured meshes
• Asymptotic properties of sample quantiles of discrete distributions
• A PDE-based approach to nondominated sorting
• Title not available (Why is that?)
• A Hamilton-Jacobi equation for the continuum limit of nondominated sorting
• The convex floating body.
• Fast Marching Methods
• Optimal control and viscosity solutions of Hamilton-Jacobi-Bellman equations
• Approximation schemes for viscosity solutions of Hamilton-Jacobi equations
• Two Approximations of Solutions of Hamilton-Jacobi Equations
• Uniqueness and error analysis for Hamilton-Jacobi equations with discontinuities
• Asymptotics for multivariate trimming
• On the Stahel-Donoho estimator and depth-weighted means of multivariate data.
• Asymptotics for the Tukey depth process, with an application to a multivariate trimmed mean
• Fast Computation of Tukey Trimmed Regions and Median in Dimension p > 2
• An introduction to viscosity solutions for fully nonlinear PDE with applications to calculus of variations in \(L^\infty\)
• Monge-Kantorovich depth, quantiles, ranks and signs
• Title not available (Why is that?)
• Approximate computation of projection depths
• Halfspace depth and floating body
• Absolute approximation of Tukey depth: theory and experiments
• Approximate calculation of Tukey's depth and median with high-dimensional data
• Hamilton–Jacobi Equations
• On smoothness of Tukey depth contours
• The limit shape of convex hull peeling
• Rates of convergence for the continuum limit of nondominated sorting

Cited In (4)

• Boundary estimation from point clouds: algorithms, guarantees and applications
• Eikonal depth: an optimal control approach to statistical depths
• Strong limit theorems for empirical halfspace depth trimmed regions
• Monotone discretizations of levelset convex geometric PDEs

This page was built for publication: Tukey depths and Hamilton-Jacobi differential equations

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5075722)