Aequa: Fair Model Rewards in Collaborative Learning via Slimmable Networks

2025-02-07

AI Summary

To address unfair reward allocation in collaborative learning, this paper proposes a contribution-driven fair allocation framework based on slimmable neural networks (SNNs). Methodologically, it constructs a shared global model with adjustable width and smoothly degraded performance as width decreases; then designs a post-training width-allocation algorithm that dynamically assigns personalized submodels—each with an optimal width—according to each participant’s contribution. The key innovation lies in the first integration of SNNs into collaborative learning incentive mechanisms, establishing a theoretically grounded mapping between model width and contribution, thereby enabling fine-grained, provably convergent fairness guarantees. Extensive experiments across multiple datasets and architectures validate the consistency between fairness of allocation and model utility. Theoretical analysis ensures algorithmic convergence and supports real-time reward distribution during training.

Technology Category

• Machine Learning: Online Learning & Bandits
• Search and Optimization: Learning to Search
• Game Theory and Economic Paradigms: Fair Division

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Economics, Online Markets and Human Computation: Fairness and ethical considerations in crowd work and in human-in-the-loop AI systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Collaborative learning enables multiple participants to learn a single global model by exchanging focused updates instead of sharing data. One of the core challenges in collaborative learning is ensuring that participants are rewarded fairly for their contributions, which entails two key sub-problems: contribution assessment and reward allocation. This work focuses on fair reward allocation, where the participants are incentivized through model rewards - differentiated final models whose performance is commensurate with the contribution. In this work, we leverage the concept of slimmable neural networks to collaboratively learn a shared global model whose performance degrades gracefully with a reduction in model width. We also propose a post-training fair allocation algorithm that determines the model width for each participant based on their contributions. We theoretically study the convergence of our proposed approach and empirically validate it using extensive experiments on different datasets and architectures. We also extend our approach to enable training-time model reward allocation.

Problem

• Fair reward allocation in collaborative learning
• Slimmable networks for model differentiation
• Contribution-based model width determination

Innovation

• Slimmable neural networks
• Post-training fair allocation
• Training-time model reward