Saved in:
Bibliographic Details
Main Authors: Domingos, João, Xavier, João
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2405.09436
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913351648411648
author Domingos, João
Xavier, João
author_facet Domingos, João
Xavier, João
contents We consider the problem of robustly fitting a model to data that includes outliers by formulating a percentile optimization problem. This problem is non-smooth and non-convex, hence hard to solve. We derive properties that the minimizers of such problems must satisfy. These properties lead to methods that solve the percentile formulation both for general residuals and for convex residuals. The methods fit the model to subsets of the data, and then extract the solution of the percentile formulation from these partial fits. As illustrative simulations show, such methods endure higher outlier percentages, when compared with standard robust estimates. Additionally, the derived properties provide a broader and alternative theoretical validation for existing robust methods, whose validity was previously limited to specific forms of the residuals.
format Preprint
id arxiv_https___arxiv_org_abs_2405_09436
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Outlier-resilient model fitting via percentile losses: Methods for general and convex residuals
Domingos, João
Xavier, João
Signal Processing
We consider the problem of robustly fitting a model to data that includes outliers by formulating a percentile optimization problem. This problem is non-smooth and non-convex, hence hard to solve. We derive properties that the minimizers of such problems must satisfy. These properties lead to methods that solve the percentile formulation both for general residuals and for convex residuals. The methods fit the model to subsets of the data, and then extract the solution of the percentile formulation from these partial fits. As illustrative simulations show, such methods endure higher outlier percentages, when compared with standard robust estimates. Additionally, the derived properties provide a broader and alternative theoretical validation for existing robust methods, whose validity was previously limited to specific forms of the residuals.
title Outlier-resilient model fitting via percentile losses: Methods for general and convex residuals
topic Signal Processing
url https://arxiv.org/abs/2405.09436