Sparse MoEs meet Efficient Ensembles

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.

59 pages, 26 figures, 36 tables. Accepted at TMLR

References in corpus (15)

• Distilling the Knowledge in a Neural Network
• Learning Transferable Visual Models From Natural Language Supervision
• Language Models are Few-Shot Learners
• Popular Ensemble Methods: An Empirical Study
• On Calibration of Modern Neural Networks
• Do ImageNet Classifiers Generalize to ImageNet?
• Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity
• Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer
• Carbon Emissions and Large Neural Network Training
• Exploring the Limits of Out-of-Distribution Detection
• BatchEnsemble: An Alternative Approach to Efficient Ensemble and Lifelong Learning
• Revisiting the Calibration of Modern Neural Networks
• Scaling Vision with Sparse Mixture of Experts
• Deep Ensembles for Low-Data Transfer Learning
• Towards Robust Vision Transformer