Sparse MoEs meet Efficient Ensembles

TL;DR

This paper introduces E$^3$, an efficient ensemble of sparse MoEs that combines the strengths of neural network ensembles and sparse MoEs, achieving better accuracy and uncertainty estimates with fewer FLOPs.

Contribution

The paper presents E$^3$, a scalable and simple ensemble method that integrates sparse MoEs and ensembles, improving performance and efficiency over existing models.

Findings

E$^3$ reduces FLOPs by up to 45% compared to deep ensembles.

E$^3$ improves accuracy, log-likelihood, and uncertainty estimates.

E$^3$ scales effectively to models with up to 2.7B parameters.

Abstract

Machine learning models based on the aggregated outputs of submodels, either at the activation or prediction levels, often exhibit strong performance compared to individual models. We study the interplay of two popular classes of such models: ensembles of neural networks and sparse mixture of experts (sparse MoEs). First, we show that the two approaches have complementary features whose combination is beneficial. This includes a comprehensive evaluation of sparse MoEs in uncertainty related benchmarks. Then, we present Efficient Ensemble of Experts (E$^3$), a scalable and simple ensemble of sparse MoEs that takes the best of both classes of models, while using up to 45% fewer FLOPs than a deep ensemble. Extensive experiments demonstrate the accuracy, log-likelihood, few-shot learning, robustness, and uncertainty improvements of E$^3$ over several challenging vision Transformer-based baselines. E$^3$ not only preserves its efficiency while scaling to models with up to 2.7B parameters, but also provides better predictive performance and uncertainty estimates for larger models.