Robustifying likelihoods by optimistically re-weighting data

Author name not available (Why is that?)

Publication date: 18 March 2023

Abstract: Likelihood-based inferences have been remarkably successful in wide-spanning application areas. However, even after due diligence in selecting a good model for the data at hand, there is inevitably some amount of model misspecification: outliers, data contamination or inappropriate parametric assumptions such as Gaussianity mean that most models are at best rough approximations of reality. A significant practical concern is that for certain inferences, even small amounts of model misspecification may have a substantial impact; a problem we refer to as brittleness. This article attempts to address the brittleness problem in likelihood-based inferences by choosing the most model friendly data generating process in a discrepancy-based neighbourhood of the empirical measure. This leads to a new Optimistically Weighted Likelihood (OWL), which robustifies the original likelihood by formally accounting for a small amount of model misspecification. Focusing on total variation (TV) neighborhoods, we study theoretical properties, develop inference algorithms and illustrate the methodology in applications to mixture models and regression.

Has companion code repository: https://github.com/cjtosh/owl

This page was built for publication: Robustifying likelihoods by optimistically re-weighting data

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6430053)