Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Portfolio-Based Incentive Mechanism Design for Cross-Device Federated Learning

Jiaxi Yang, Sheng Cao, Cuifang Zhao, Weina Niu, Li-Chuan Tsai

TL;DR

This work tackles incentive design for cross-device federated learning under stochastic and correlated client capacities by introducing a portfolio-based mechanism (FL-IMP) guided by Markowitz portfolio theory. It integrates a multi-level FL architecture (IoTs, EBSs, MBSs) with an agent-based model (FL-ABM) to realistically simulate uncertainty and inter-client correlations, enabling budget-aware reward allocation. The key contributions include applying portfolio optimization to FL resource allocation, deriving an efficient frontier characterization, and demonstrating superior performance over auction, random, and greedy baselines across MNIST, Fashion-MNIST, and CIFAR-10 datasets with both IID and Non-IID partitions. The approach offers a principled, scalable method for managing uncertainty and correlation in cross-device FL, with practical implications for resource-constrained, privacy-preserving collaborative learning systems.

Abstract

In recent years, there has been a significant increase in attention towards designing incentive mechanisms for federated learning (FL). Tremendous existing studies attempt to design the solutions using various approaches (e.g., game theory, reinforcement learning) under different settings. Yet the design of incentive mechanism could be significantly biased in that clients' performance in many applications is stochastic and hard to estimate. Properly handling this stochasticity motivates this research, as it is not well addressed in pioneering literature. In this paper, we focus on cross-device FL and propose a multi-level FL architecture under the real scenarios. Considering the two properties of clients' situations: uncertainty, correlation, we propose FL Incentive Mechanism based on Portfolio theory (FL-IMP). As far as we are aware, this is the pioneering application of portfolio theory to incentive mechanism design aimed at resolving FL resource allocation problem. In order to more accurately reflect practical FL scenarios, we introduce the Federated Learning Agent-Based Model (FL-ABM) as a means of simulating autonomous clients. FL-ABM enables us to gain a deeper understanding of the factors that influence the system's outcomes. Experimental evaluations of our approach have extensively validated its effectiveness and superior performance in comparison to the benchmark methods.

Portfolio-Based Incentive Mechanism Design for Cross-Device Federated Learning

TL;DR

This work tackles incentive design for cross-device federated learning under stochastic and correlated client capacities by introducing a portfolio-based mechanism (FL-IMP) guided by Markowitz portfolio theory. It integrates a multi-level FL architecture (IoTs, EBSs, MBSs) with an agent-based model (FL-ABM) to realistically simulate uncertainty and inter-client correlations, enabling budget-aware reward allocation. The key contributions include applying portfolio optimization to FL resource allocation, deriving an efficient frontier characterization, and demonstrating superior performance over auction, random, and greedy baselines across MNIST, Fashion-MNIST, and CIFAR-10 datasets with both IID and Non-IID partitions. The approach offers a principled, scalable method for managing uncertainty and correlation in cross-device FL, with practical implications for resource-constrained, privacy-preserving collaborative learning systems.

Abstract

In recent years, there has been a significant increase in attention towards designing incentive mechanisms for federated learning (FL). Tremendous existing studies attempt to design the solutions using various approaches (e.g., game theory, reinforcement learning) under different settings. Yet the design of incentive mechanism could be significantly biased in that clients' performance in many applications is stochastic and hard to estimate. Properly handling this stochasticity motivates this research, as it is not well addressed in pioneering literature. In this paper, we focus on cross-device FL and propose a multi-level FL architecture under the real scenarios. Considering the two properties of clients' situations: uncertainty, correlation, we propose FL Incentive Mechanism based on Portfolio theory (FL-IMP). As far as we are aware, this is the pioneering application of portfolio theory to incentive mechanism design aimed at resolving FL resource allocation problem. In order to more accurately reflect practical FL scenarios, we introduce the Federated Learning Agent-Based Model (FL-ABM) as a means of simulating autonomous clients. FL-ABM enables us to gain a deeper understanding of the factors that influence the system's outcomes. Experimental evaluations of our approach have extensively validated its effectiveness and superior performance in comparison to the benchmark methods.
Paper Structure (11 sections, 14 equations, 3 figures, 3 tables, 1 algorithm)

This paper contains 11 sections, 14 equations, 3 figures, 3 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Multi-level Federated Learning Architecture
  4. Problem Formulation
  5. Portfolio based Incentive Mechanism of FL
  6. Experiments
  7. Experiment Settings
  8. Benchmarks
  9. Experiment Results
  10. Conclusion
  11. Acknowledgement

Figures (3)

  • Figure 1: Overview of Multi-level Federated Learning Architecture. The three layers from top to bottom consist of aggregation servers (MBSs), edge computing devices (EBSs), and clients (IoTs).
  • Figure 2: Test accuracy during the training process on MNIST, Fashion-MNIST and CIFAR-10, arranged in ascending order from top to bottom
  • Figure 3: The utility with FedAvg and FedProx on IID and Non-IID Dataset.