Model Selection by Loss Rank for Classification and Unsupervised Learning

TL;DR

This paper extends the loss rank principle (LoRP) to classification and unsupervised learning, providing theoretical insights and simulation results for model selection in these contexts.

Contribution

It develops the LoRP framework for classification and unsupervised learning, expanding its applicability and analyzing its theoretical properties.

Findings

LoRP is effective for classification model selection.

Theoretical properties of LoRP are established.

Simulation studies demonstrate its practical utility.

Abstract

Hutter (2007) recently introduced the loss rank principle (LoRP) as a generalpurpose principle for model selection. The LoRP enjoys many attractive properties and deserves further investigations. The LoRP has been well-studied for regression framework in Hutter and Tran (2010). In this paper, we study the LoRP for classification framework, and develop it further for model selection problems in unsupervised learning where the main interest is to describe the associations between input measurements, like cluster analysis or graphical modelling. Theoretical properties and simulation studies are presented.