Don't Forget What I did?: Assessing Client Contributions in Federated Learning

2024-03-11

arXiv.org

AI Summary

In federated learning, dynamic sparse participation—where only a subset of clients is active per round—renders existing Shapley-value-based contribution evaluation methods ineffective, as they neglect both historical client behavior and non-i.i.d. data distributions. To address this, we propose FLContrib, a history-aware game-theoretic framework. First, it introduces a timeline-based contribution model that jointly incorporates clients’ historical participation patterns and local data characteristics. Second, it designs a bilateral fair scheduling algorithm to enable efficient sparse approximation of Shapley values. Third, it integrates a lightweight data poisoning detection mechanism. Extensive evaluations on multiple benchmark datasets demonstrate that FLContrib maintains high assessment accuracy (relative error < 5.3%), reduces computational overhead to 1/8 that of conventional methods, and achieves a false positive rate < 3.2% for poisoned client identification.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization
• Multiagent Systems: Adversarial Agents

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Security and Privacy: Large-scale security measurements
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Federated Learning (FL) is a collaborative machine learning (ML) approach, where multiple clients participate in training an ML model without exposing the private data. Fair and accurate assessment of client contributions is an important problem in FL to facilitate incentive allocation and encouraging diverse clients to participate in a unified model training. Existing methods for assessing client contribution adopts co-operative game-theoretic concepts, such as Shapley values, but under simplified assumptions. In this paper, we propose a history-aware game-theoretic framework, called FLContrib, to assess client contributions when a subset of (potentially non-i.i.d.) clients participate in each epoch of FL training. By exploiting the FL training process and linearity of Shapley value, we develop FLContrib that yields a historical timeline of client contributions as FL training progresses over epochs. Additionally, to assess client contribution under limited computational budget, we propose a scheduling procedure that considers a two-sided fairness criteria to perform expensive Shapley value computation only in a subset of training epochs. In experiments, we demonstrate a controlled trade-off between the correctness and efficiency of client contributions assessed via FLContrib. To demonstrate the benefits of history-aware client contributions, we apply FLContrib to detect dishonest clients conducting data poisoning in FL training.

Problem

• Assessing client contributions in federated learning with dynamic participation
• Proposing a fair method to compute historical Shapley values efficiently
• Detecting dishonest clients using historical contribution timelines

Innovation

• History-aware client contribution assessment framework
• Dynamic client participation with Markovian training process
• Two-sided fairness criteria for Shapley computation scheduling