Understanding the Role of Functional Diversity in Weight-Ensembling with Ingredient Selection and Multidimensional Scaling

TL;DR

This paper investigates how functional diversity among neural network models influences the effectiveness of weight-ensembles, introducing new methods and visualization tools to analyze model selection and diversity effects.

Contribution

It proposes two novel weight-ensembling methods and a visualization tool to better understand the role of functional diversity in ensemble performance.

Findings

High diversity enhances weight-ensembling effectiveness.

Sampling positionally distinct models can improve ensemble performance.

Diversity alone does not fully account for accuracy gains.

Abstract

Weight-ensembles are formed when the parameters of multiple neural networks are directly averaged into a single model. They have demonstrated generalization capability in-distribution (ID) and out-of-distribution (OOD) which is not completely understood, though they are thought to successfully exploit functional diversity allotted by each distinct model. Given a collection of models, it is also unclear which combination leads to the optimal weight-ensemble; the SOTA is a linear-time ``greedy" method. We introduce two novel weight-ensembling approaches to study the link between performance dynamics and the nature of how each method decides to use apply the functionally diverse components, akin to diversity-encouragement in the prediction-ensemble literature. We develop a visualization tool to explain how each algorithm explores various domains defined via pairwise-distances to further investigate selection and algorithms' convergence. Empirical analyses shed perspectives which reinforce how high-diversity enhances weight-ensembling while qualifying the extent to which diversity alone improves accuracy. We also demonstrate that sampling positionally distinct models can contribute just as meaningfully to improvements in a weight-ensemble.