Vector quantile regression and optimal transport, from theory to numerics

  title={Vector quantile regression and optimal transport, from theory to numerics},
  author={Guillaume Carlier and Victor Chernozhukov and Gwendoline De Bie and Alfred Galichon},
  journal={Empirical Economics},
In this paper, we first revisit the Koenker and Bassett variational approach to (univariate) quantile regression, emphasizing its link with latent factor representations and correlation maximization problems. We then review the multivariate extension due to Carlier et al. (Ann Statist 44(3):1165–92, 2016,; J Multivariate Anal 161:96–102, 2017) which relates vector quantile regression to an optimal transport problem with mean independence constraints. We introduce an entropic regularization of… 

Deep Non-Crossing Quantiles through the Partial Derivative

A generic deep learning algorithm for predict-ing an arbitrary number of quantiles that ensures the quantile monotonicity constraint up to the machine precision and maintains its modelling performance with respect to alternative models is proposed.

Calibrated Multiple-Output Quantile Regression with Representation Learning

An extension of conformal prediction to the multivariate response setting that modifies any method to return sets with a pre-specified coverage level is proposed, which is theoretically guaranteed in the finite-sample case for any distribution.

Fast Nonlinear Vector Quantile Regression

Quantile regression (QR) is a powerful tool for estimating one or more conditional quantiles of a target variable Y given explanatory features X. A limitation of QR is that it is only defined for

Limit Theorems for Entropic Optimal Transport Maps and the Sinkhorn Divergence

. We study limit theorems for entropic optimal transport (EOT) maps, dual potentials, and the Sinkhorn divergence. The key technical tool we use is a first and second order Hadamard differentiability

Economic applications of quantile regression 2.0

Twenty years ago, Empirical Economics was instrumental in supporting a conference at the University of Konstanz on economic applications of quantile regression that attracted a number of influential

The unreasonable effectiveness of optimal transport in economics

A brief review of the basics of the theory of optimal transport, its applications to economics, and some extensions are offered.

Decorrelation with conditional normalizing flows

A normalizing conditioned on the protected attributes can be used to create a decorrelated representation for any discriminant, and as a normalizing is invertible the separation power of the resulting discriminant will be unchanged at any number of protected attributes.



Vector quantile regression beyond the specified case

A Smoothed Dual Approach for Variational Wasserstein Problems

The dual formulation of Wasserstein variational problems introduced recently can be regularized using an entropic smoothing, which leads to smooth, differentiable, convex optimization problems that are simpler to implement and numerically more stable.

Vector quantile regression: An optimal transport approach

We propose a notion of conditional vector quantile function and a vector quantile regression. A conditional vector quantile function (CVQF) of a random vector Y, taking values in Rd given covariates

Stochastic Optimization for Large-scale Optimal Transport

A new class of stochastic optimization algorithms to cope with large-scale problems routinely encountered in machine learning applications, based on entropic regularization of the primal OT problem, which results in a smooth dual optimization optimization which can be addressed with algorithms that have a provably faster convergence.

Polar Factorization and Monotone Rearrangement of Vector-Valued Functions

Given a probability space (X, p) and a bounded domain R in Rd equipped with the Lebesgue measure 1 . I (normalized so that 10 I = I ), it is shown (under additional technical assumptions on X and Q)

The regularity of mappings with a convex potential

In this work, we apply the techniques developed in [Cl] to the problem of mappings with a convex potential between domains. That is, given two bounded domains Q, Q2 of Rn and two nonnegative real

Existence and uniqueness of monotone measure-preserving maps

Introduction. Given a pair of Borel probability measures # and v on Rd, it is natural to ask whether v can be obtained from # by redistributing its mass in a canonical way. In the case of the line d

A Fast Iterative Shrinkage-Thresholding Algorithm for Linear Inverse Problems

A new fast iterative shrinkage-thresholding algorithm (FISTA) which preserves the computational simplicity of ISTA but with a global rate of convergence which is proven to be significantly better, both theoretically and practically.

Regression Quantiles