Shuffled total least squares
Statistics Theory
2022-09-05 v1 Statistics Theory
Abstract
Linear regression with shuffled labels and with a noisy latent design matrix arises in many correspondence recovery problems. We propose a total least-squares approach to the problem of estimating the underlying true permutation and provide an upper bound to the normalized Procrustes quadratic loss of the estimator. We also provide an iterative algorithm to approximate the estimator and demonstrate its performance on simulated data.
Cite
@article{arxiv.2209.01066,
title = {Shuffled total least squares},
author = {Qian Wang and Daniel Sussman},
journal= {arXiv preprint arXiv:2209.01066},
year = {2022}
}