Omni-Router: Sharing Routing Decisions in Sparse Mixture-of-Experts for Speech Recognition
This work addresses inefficiencies in ASR systems for users needing robust performance across diverse datasets, though it is incremental as it builds on existing MoE methods.
The paper tackles the problem of inefficient expert routing in mixture-of-experts architectures for speech recognition by introducing a shared router across layers, resulting in reduced word error rates by 11.2% and 8.2% compared to dense and Switch Transformer models.
Mixture-of-experts (MoE) architectures have expanded from language modeling to automatic speech recognition (ASR). Traditional MoE methods, such as the Switch Transformer, route experts independently within each layer. Our analysis reveals that routers in most layers make expert choices that are not strongly correlated with the choices of the routers in other layers. To increase the cooperation between experts in different layers and encourage greater specialization, we use a shared router across different MoE layers. We call this model Omni-router Transformer. Extensive experiments on a large-scale pseudo-labeled dataset and evaluations across 10 diverse, out-of-domain ASR benchmarks demonstrate that the Omni-router Transformer is able to achieve lower training loss and consistently outperform dense and Switch Transformer models, reducing average word error rates by 11.2% and 8.2%, respectively, while providing structured expert usage and improved robustness to diverse data.