Model marketplaces coordinate experts with cost aware mixtures
Cost-Aware Mixture-of-Experts Coordination for Model Markets
DatabasesMachine Learning
Summary
Model marketplaces usually pick one fixed model for a task, missing chances to combine strengths of different models. The authors propose using a system that blends multiple experts—each with different skills and costs—into one service, balancing prediction quality and expenses. They design rules for how experts share earnings based on their contribution and cost, ensuring fairness and efficiency. Tests show this approach improves overall value and controls costs across diverse tasks without losing accuracy.
What this means in practice
- •For machine learning platform developers: Implement composite services that combine multiple heterogeneous models with cost-sensitive coordination to improve overall performance and reduce latency.
- •For cloud service operators: Deploy model markets that allocate computational resources efficiently by factoring in expert execution costs while maintaining high predictive accuracy.
Authors
Yizhou Ma, Wenbo Wu, Xikun Jiang, Zhuoqin Yang, Luis-Daniel Ibáñez
Abstract
Existing model marketplaces typically trade and select individual models as indivisible units, limiting their ability to exploit complementarities among heterogeneous experts. This paper proposes an MoE-based model market framework that lifts Mixture-of-Experts from a model-level learning architecture to a market-level coordination mechanism. In this framework, brokers use gating networks to coordinate multiple heterogeneous experts and deliver a composite model service. We formalize the market participants, service workflow, expert cost structure, and a welfare objective that combines predictive utility with heterogeneous execution costs. We then derive a cost-aware gating mechanism and market-aware training objective, and introduce a cost-adjusted revenue allocation rule that distributes residual revenue according to realized expert participation and execution cost. We also establish basic theoretical properties of the allocation rule, including budget balance, participation monotonicity, and cost sensitivity. Experiments over five random seeds on fifteen tabular and image benchmarks use independently trained and frozen neural and tree-based experts together with latency-derived execution costs. MoE Market achieves the highest mean welfare on all fifteen datasets and a lower mean expected cost than Standard MoE in every case, while maintaining competitive predictive performance. The allocation experiments further demonstrate systematic sensitivity to expert participation and cost, together with substantially lower computational overhead than exact Shapley allocation. These results suggest that MoE can serve as a market-level coordination principle for collaborative, cost-aware, and economically grounded model marketplaces.