Multilinear mixture of experts: scalable expert specialization through factorization
Author(s)
Type
Conference Paper
Abstract
The Mixture of Experts (MoE) paradigm provides a powerful way to decompose dense layers into smaller, modular computations often more amenable to human interpretation, debugging, and editability. However, a major challenge lies in the computational cost of scaling the number of experts high enough to achieve fine-grained specialization. In this paper, we propose the Multilinear Mixture of Experts (µMoE) layer to address this, focusing on vision models. µMoE layers enable scalable expert specialization by performing an implicit computation on prohibitively large weight tensors entirely in factorized form. Consequently, µMoEs (1) avoid the restrictively high inference-time costs of dense MoEs, yet (2) do not inherit the training issues of the popular sparse MoEs' discrete (non-differentiable) expert routing. We present both qualitative and quantitative evidence that scaling µMoE layers when fine-tuning foundation models for vision tasks leads to more specialized experts at the class-level, further enabling manual bias correction in CelebA attribute classification. Finally, we show qualitative results demonstrating the expert specialism achieved when pre-training large GPT2 and MLP-Mixer models with parameter-matched µMoE blocks at every layer, maintaining comparable accuracy. Our code is available at: https://github.com/james-oldfield/muMoE.
Date Issued
2025-02-01
Date Acceptance
2024-12-01
Citation
Advances in Neural Information Processing Systems, 2025, 37, pp.53022-53063
ISBN
9798331314385
ISSN
1049-5258
Publisher
Neural Information Processing Systems Foundation, Inc. (NeurIPS)
Start Page
53022
End Page
53063
Journal / Book Title
Advances in Neural Information Processing Systems
Volume
37
Copyright Statement
© 2025 Neural Information Processing Systems Foundation, Inc. (NeurIPS).
Source
NeurIPS 2024
Publication Status
Published
Start Date
2024-12-10
Finish Date
2024-12-15
Coverage Spatial
Vancouver, Canada