Uncoupled isotonic regression via minimum Wasserstein deconvolution

DOI10.1093/IMAIAI/IAZ006MaRDI QIDQ5006508zbMATH OpenOpenAlexWikidataFDO

Authors Philippe Rigollet, Jonathan Weed

Publication date 16 August 2021

Published in Information and Inference: A Journal of the IMA (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1806.10648, https://hdl.handle.net/1721.1/126717

zbMATH Keywords

couplingdeconvolutionisotonic regressionminimum Kantorovich distance estimationmoment matching

Mathematics Subject Classification ID

Nonparametric regression and quantile regression (62G08) General nonlinear regression (62J02)

Abstract: Isotonic regression is a standard problem in shape-constrained estimation where the goal is to estimate an unknown nondecreasing regression function $f$ from independent pairs $(x_i,y_i)$ where $mathbbE[y_i]=f(x_i),i=1,ldotsn$. While this problem is well understood both statistically and computationally, much less is known about its uncoupled counterpart where one is given only the unordered sets $x_1,ldots,x_n$ and $y_1,ldots,y_n$. In this work, we leverage tools from optimal transport theory to derive minimax rates under weak moments conditions on $y_i$ and to give an efficient algorithm achieving optimal rates. Both upper and lower bounds employ moment-matching arguments that are also pertinent to learning mixtures of distributions and deconvolution.

Recommendations

• Isotonic regression in general dimensions
• An O(n^2) algorithm for isotonic regression
• Non-convex isotonic regression via the Myersonian approach
• Isotonic regression in multi-dimensional spaces and graphs
• scientific article; zbMATH DE number 1267279

Cited in

(23)

• Minimax estimation of smooth optimal transport maps
• Isotonic regression with unknown permutations: statistics, computation and adaptation
• Multivariate ranks and quantiles using optimal transport: consistency, rates and nonparametric testing
• Minimax estimation of smooth densities in Wasserstein distance
• Worst-case versus average-case design for estimation from partial pairwise comparisons
• Estimation of Wasserstein distances in the spiked transport model
• A Pseudo-Likelihood Approach to Linear Regression With Partially Shuffled Data
• Two-stage approach to multivariate linear regression with sparsely mismatched data
• Unlinked monotone regression
• Uncoupled Isotonic Regression with Discrete Errors
• Optimal Permutation Recovery in Permuted Monotone Matrix Model
• Subgraph distributions in dense random regular graphs
• Recovering Latent Variables by Matching
• An Improved Central Limit Theorem and Fast Convergence Rates for Entropic Transportation Costs
• Projection-based techniques for high-dimensional optimal transport problems
• Regression with linked datasets subject to linkage error
• Matching a discrete distribution by Poisson matching quantiles estimation
• Method of moments for estimation of noisy curves
• Statistical Wasserstein distance with rank regularization and spiked structure
• Optimal transport map estimation in general function spaces
• Nearly minimax optimal Wasserstein conditional independence testing
• On the sample complexity of entropic optimal transport
• Minimax optimal rates of convergence in monotone shuffled and unlinked regression models under vanishing noise

This page was built for publication: Uncoupled isotonic regression via minimum Wasserstein deconvolution

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5006508)