Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

martFL: Enabling Utility-Driven Data Marketplace with a Robust and Verifiable Federated Learning Architecture

Qi Li, Zhuotao Liu, Qi Li, Ke Xu

TL;DR

martFL tackles privacy-preserving, utility-driven data marketplaces by introducing a quality-aware model evaluation that privately assesses all DP updates and a verifiable transaction protocol that proves, in zero-knowledge, that aggregation followed the committed weights. The approach mitigates the inclusiveness-robustness tradeoff inherent in prior FL while enabling fair reward distribution via smart contracts, without revealing private data or evaluation methods. Empirical results show up to 25% accuracy gains and up to 64% reductions in data acquisition cost, along with resilience to untargeted, targeted, and Sybil attacks across four datasets. The combination of CKKS-based privacy, zk-SNARK verification, and on-chain enforcement offers a practical, scalable path to secure data marketplaces with robust and verifiable FL pipelines.

Abstract

The development of machine learning models requires a large amount of training data. Data marketplaces are essential for trading high-quality, private-domain data not publicly available online. However, due to growing data privacy concerns, direct data exchange is inappropriate. Federated Learning (FL) is a distributed machine learning paradigm that exchanges data utilities (in form of local models or gradients) among multiple parties without directly sharing the raw data. However, several challenges exist when applying existing FL architectures to construct a data marketplace: (i) In existing FL architectures, Data Acquirers (DAs) cannot privately evaluate local models from Data Providers (DPs) prior to trading; (ii) Model aggregation protocols in existing FL designs struggle to exclude malicious DPs without "overfitting" to the DA's (possibly biased) root dataset; (iii) Prior FL designs lack a proper billing mechanism to enforce the DA to fairly allocate the reward according to contributions made by different DPs. To address above challenges, we propose martFL, the first federated learning architecture that is specifically designed to enable a secure utility-driven data marketplace. At a high level, martFL is powered by two innovative designs: (i) a quality-aware model aggregation protocol that achieves robust local model aggregation even when the DA's root dataset is biased; (ii) a verifiable data transaction protocol that enables the DA to prove, both succinctly and in zero-knowledge, that it has faithfully aggregates the local models submitted by different DPs according to the committed aggregation weights, based on which the DPs can unambiguously claim the corresponding reward. We implement a prototype of martFL and evaluate it extensively over various tasks. The results show that martFL can improve the model accuracy by up to 25% while saving up to 64% data acquisition cost.

martFL: Enabling Utility-Driven Data Marketplace with a Robust and Verifiable Federated Learning Architecture

TL;DR

martFL tackles privacy-preserving, utility-driven data marketplaces by introducing a quality-aware model evaluation that privately assesses all DP updates and a verifiable transaction protocol that proves, in zero-knowledge, that aggregation followed the committed weights. The approach mitigates the inclusiveness-robustness tradeoff inherent in prior FL while enabling fair reward distribution via smart contracts, without revealing private data or evaluation methods. Empirical results show up to 25% accuracy gains and up to 64% reductions in data acquisition cost, along with resilience to untargeted, targeted, and Sybil attacks across four datasets. The combination of CKKS-based privacy, zk-SNARK verification, and on-chain enforcement offers a practical, scalable path to secure data marketplaces with robust and verifiable FL pipelines.

Abstract

The development of machine learning models requires a large amount of training data. Data marketplaces are essential for trading high-quality, private-domain data not publicly available online. However, due to growing data privacy concerns, direct data exchange is inappropriate. Federated Learning (FL) is a distributed machine learning paradigm that exchanges data utilities (in form of local models or gradients) among multiple parties without directly sharing the raw data. However, several challenges exist when applying existing FL architectures to construct a data marketplace: (i) In existing FL architectures, Data Acquirers (DAs) cannot privately evaluate local models from Data Providers (DPs) prior to trading; (ii) Model aggregation protocols in existing FL designs struggle to exclude malicious DPs without "overfitting" to the DA's (possibly biased) root dataset; (iii) Prior FL designs lack a proper billing mechanism to enforce the DA to fairly allocate the reward according to contributions made by different DPs. To address above challenges, we propose martFL, the first federated learning architecture that is specifically designed to enable a secure utility-driven data marketplace. At a high level, martFL is powered by two innovative designs: (i) a quality-aware model aggregation protocol that achieves robust local model aggregation even when the DA's root dataset is biased; (ii) a verifiable data transaction protocol that enables the DA to prove, both succinctly and in zero-knowledge, that it has faithfully aggregates the local models submitted by different DPs according to the committed aggregation weights, based on which the DPs can unambiguously claim the corresponding reward. We implement a prototype of martFL and evaluate it extensively over various tasks. The results show that martFL can improve the model accuracy by up to 25% while saving up to 64% data acquisition cost.
Paper Structure (37 sections, 6 equations, 11 figures, 8 tables, 2 algorithms)

This paper contains 37 sections, 6 equations, 11 figures, 8 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Background and Motivation
  3. Data Marketplace
  4. Federated Learning and Its Robustness
  5. Zero-Knowledge Proofs and Verifiable Machine Learning
  6. Motivation
  7. Assumptions and Threat Model
  8. martFL Overview
  9. Quality-Aware Model Evaluations
  10. Key Observations
  11. Model Aggregation Protocol
  12. Hierarchical Clustering for Outlier Removal
  13. Dynamic Baseline Adjustment
  14. The Integrated Training Process
  15. Verifiable Transaction Protocol
  16. ...and 22 more sections

Figures (11)

  • Figure 1: The architecture comparison between the vanilla FL and martFL.
  • Figure 2: The tradeoff between robustness and inclusiveness in prior robust FL approaches.
  • Figure 3: Three different cases of distribution of scores.
  • Figure 4: The circuit design for the proving scheme in martFL.
  • Figure 5: The inclusiveness analysis when the DA posseses a biased root dataset.
  • ...and 6 more figures

Theorems & Definitions (1)

  • definition 1