Aequa: Fair Model Rewards in Collaborative Learning via Slimmable Networks

TL;DR

This paper introduces Aequa, a method for fair reward allocation in collaborative learning using slimmable networks, ensuring contributions are fairly reflected in the final model performance.

Contribution

It proposes a novel post-training and training-time reward allocation approach leveraging slimmable neural networks for fair contribution assessment.

Findings

The approach achieves fair reward distribution aligning with contributions.

The method demonstrates convergence and effectiveness across various datasets.

Empirical validation confirms improved fairness in collaborative learning settings.

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.