CARE: Compatibility-Aware Incentive Mechanisms for Federated Learning with Budgeted Requesters

TL;DR

This work addresses incentive design for federated learning with multiple budgeted requesters and incompatible workers. It introduces CARE, comprising CARE-CO for cooperative budgets and CARE-NO for non-cooperative budgets, using a Max-Flow reduction and a virtual-price based PEA mechanism to handle compatibility constraints and private costs. The mechanisms guarantee individual rationality, truthfulness, budget feasibility, and offer approximation guarantees, demonstrated by experiments on FMNIST and CIFAR-10 that show substantial improvements in worker reputation and global model accuracy. The results highlight the practical impact of incorporating worker compatibility and budget considerations into FL incentive design, enabling more efficient and effective collaborations across requesting entities and heterogeneous workers.

Abstract

Federated learning (FL) is a promising approach that allows requesters (\eg, servers) to obtain local training models from workers (e.g., clients). Since workers are typically unwilling to provide training services/models freely and voluntarily, many incentive mechanisms in FL are designed to incentivize participation by offering monetary rewards from requesters. However, existing studies neglect two crucial aspects of real-world FL scenarios. First, workers can possess inherent incompatibility characteristics (e.g., communication channels and data sources), which can lead to degradation of FL efficiency (e.g., low communication efficiency and poor model generalization). Second, the requesters are budgeted, which limits the amount of workers they can hire for their tasks. In this paper, we investigate the scenario in FL where multiple budgeted requesters seek training services from incompatible workers with private training costs. We consider two settings: the cooperative budget setting where requesters cooperate to pool their budgets to improve their overall utility and the non-cooperative budget setting where each requester optimizes their utility within their own budgets. To address efficiency degradation caused by worker incompatibility, we develop novel compatibility-aware incentive mechanisms, CARE-CO and CARE-NO, for both settings to elicit true private costs and determine workers to hire for requesters and their rewards while satisfying requester budget constraints. Our mechanisms guarantee individual rationality, truthfulness, budget feasibility, and approximation performance. We conduct extensive experiments using real-world datasets to show that the proposed mechanisms significantly outperform existing baselines.

Figures (5)

• Figure 1: FL with incompatible workers and budgeted requesters.
• Figure 2: The maximum employable worker curves under the price set $R_b$, where the black lines represent requesters' employability and the blue star represents the value of $\mathcal{M}_f(r)$.
• Figure 3: $k$ workers' bids and weights when $s_i$ bids a false cost.
• Figure 4: The overall reputation of selected workers, with solid lines representing the reputation in the cooperative budget setting and dashed lines representing the reputation in the non-cooperative budget setting.
• Figure 5: The average global accuracy of the proposed mechanisms under non-IID label distribution datasets.

Theorems & Definitions (14)

• Definition 1: Sub-problem ORP
• Theorem 1
• proof : Proof Sketch
• Definition 2: Sub-problem OSP
• Theorem 2
• Lemma 1
• proof
• Lemma 2
• proof
• Lemma 3