Beyond Mixtures and Products for Ensemble Aggregation: A Likelihood Perspective on Generalized Means

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Razafindralambo, Raphaël, Sun, Rémy, Precioso, Frédéric, Garreau, Damien, Mattei, Pierre-Alexandre
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866911484538257408
author Razafindralambo, Raphaël
Sun, Rémy
Precioso, Frédéric
Garreau, Damien
Mattei, Pierre-Alexandre
author_facet Razafindralambo, Raphaël
Sun, Rémy
Precioso, Frédéric
Garreau, Damien
Mattei, Pierre-Alexandre
contents Density aggregation is a central problem in machine learning, for instance when combining predictions from a Deep Ensemble. The choice of aggregation remains an open question with two commonly proposed approaches being linear pooling (probability averaging) and geometric pooling (logit averaging). In this work, we address this question by studying the normalized generalized mean of order $r \in \mathbb{R} \cup \{-\infty,+\infty\}$ through the lens of log-likelihood, the standard evaluation criterion in machine learning. This provides a unifying aggregation formalism and shows different optimal configurations for different situations. We show that the regime $r \in [0,1]$ is the only range ensuring systematic improvements relative to individual distributions, thereby providing a principled justification for the reliability and widespread practical use of linear ($r=1$) and geometric ($r=0$) pooling. In contrast, we show that aggregation rules with $r \notin [0,1]$ may fail to provide consistent gains with explicit counterexamples. Finally, we corroborate our theoretical findings with empirical evaluations using Deep Ensembles on image and text classification benchmarks.
format Preprint
id arxiv_https___arxiv_org_abs_2603_04204
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Beyond Mixtures and Products for Ensemble Aggregation: A Likelihood Perspective on Generalized Means
Razafindralambo, Raphaël
Sun, Rémy
Precioso, Frédéric
Garreau, Damien
Mattei, Pierre-Alexandre
Machine Learning
Computer Vision and Pattern Recognition
Statistics Theory
Methodology
62-08 (Primary) 60F10 (Secondary)
G.3
Density aggregation is a central problem in machine learning, for instance when combining predictions from a Deep Ensemble. The choice of aggregation remains an open question with two commonly proposed approaches being linear pooling (probability averaging) and geometric pooling (logit averaging). In this work, we address this question by studying the normalized generalized mean of order $r \in \mathbb{R} \cup \{-\infty,+\infty\}$ through the lens of log-likelihood, the standard evaluation criterion in machine learning. This provides a unifying aggregation formalism and shows different optimal configurations for different situations. We show that the regime $r \in [0,1]$ is the only range ensuring systematic improvements relative to individual distributions, thereby providing a principled justification for the reliability and widespread practical use of linear ($r=1$) and geometric ($r=0$) pooling. In contrast, we show that aggregation rules with $r \notin [0,1]$ may fail to provide consistent gains with explicit counterexamples. Finally, we corroborate our theoretical findings with empirical evaluations using Deep Ensembles on image and text classification benchmarks.
title Beyond Mixtures and Products for Ensemble Aggregation: A Likelihood Perspective on Generalized Means
topic Machine Learning
Computer Vision and Pattern Recognition
Statistics Theory
Methodology
62-08 (Primary) 60F10 (Secondary)
G.3
url https://arxiv.org/abs/2603.04204