Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

OmniLytics+: A Secure, Efficient, and Affordable Blockchain Data Market for Machine Learning through Off-Chain Processing

Songze Li, Mingzhe Liu, Mengqi Chen

TL;DR

OmniLytics+ tackles the challenge of secure, private, and scalable data markets for ML by integrating blockchain smart contracts with off-chain ZK rollups and SNIP-based verifications. It protects both model privacy (via model perturbation) and data privacy (via secret sharing) while guarding against malicious data owners through a verifiable gradient computation workflow. Off-chain processing reduces on-chain storage and computation costs, enabling model updates with bound gas usage and scalable validation. Empirical results on Ethereum show strong privacy and robustness guarantees with substantial improvements in gas and execution time over fully on-chain or baseline rollup approaches, highlighting practical impact for data-driven ML development.

Abstract

The rapid development of large machine learning (ML) models requires a massive amount of training data, resulting in booming demands of data sharing and trading through data markets. Traditional centralized data markets suffer from low level of security, and emerging decentralized platforms are faced with efficiency and privacy challenges. In this paper, we propose OmniLytics+, the first decentralized data market, built upon blockchain and smart contract technologies, to simultaneously achieve 1) data (resp., model) privacy for the data (resp. model) owner; 2) robustness against malicious data owners; 3) efficient data validation and aggregation. Specifically, adopting the zero-knowledge (ZK) rollup paradigm, OmniLytics+ proposes to secret share encrypted local gradients, computed from the encrypted global model, with a set of untrusted off-chain servers, who collaboratively generate a ZK proof on the validity of the gradient. In this way, the storage and processing overheads are securely offloaded from blockchain verifiers, significantly improving the privacy, efficiency, and affordability over existing rollup solutions. We implement the proposed OmniLytics+ data market as an Ethereum smart contract [41]. Extensive experiments demonstrate the effectiveness of OmniLytics+ in training large ML models in presence of malicious data owner, and the substantial advantages of OmniLytics+ in gas cost and execution time over baselines.

OmniLytics+: A Secure, Efficient, and Affordable Blockchain Data Market for Machine Learning through Off-Chain Processing

TL;DR

OmniLytics+ tackles the challenge of secure, private, and scalable data markets for ML by integrating blockchain smart contracts with off-chain ZK rollups and SNIP-based verifications. It protects both model privacy (via model perturbation) and data privacy (via secret sharing) while guarding against malicious data owners through a verifiable gradient computation workflow. Off-chain processing reduces on-chain storage and computation costs, enabling model updates with bound gas usage and scalable validation. Empirical results on Ethereum show strong privacy and robustness guarantees with substantial improvements in gas and execution time over fully on-chain or baseline rollup approaches, highlighting practical impact for data-driven ML development.

Abstract

The rapid development of large machine learning (ML) models requires a massive amount of training data, resulting in booming demands of data sharing and trading through data markets. Traditional centralized data markets suffer from low level of security, and emerging decentralized platforms are faced with efficiency and privacy challenges. In this paper, we propose OmniLytics+, the first decentralized data market, built upon blockchain and smart contract technologies, to simultaneously achieve 1) data (resp., model) privacy for the data (resp. model) owner; 2) robustness against malicious data owners; 3) efficient data validation and aggregation. Specifically, adopting the zero-knowledge (ZK) rollup paradigm, OmniLytics+ proposes to secret share encrypted local gradients, computed from the encrypted global model, with a set of untrusted off-chain servers, who collaboratively generate a ZK proof on the validity of the gradient. In this way, the storage and processing overheads are securely offloaded from blockchain verifiers, significantly improving the privacy, efficiency, and affordability over existing rollup solutions. We implement the proposed OmniLytics+ data market as an Ethereum smart contract [41]. Extensive experiments demonstrate the effectiveness of OmniLytics+ in training large ML models in presence of malicious data owner, and the substantial advantages of OmniLytics+ in gas cost and execution time over baselines.
Paper Structure (24 sections, 14 equations, 7 figures, 3 tables)

This paper contains 24 sections, 14 equations, 7 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Works
  3. Federated learning on blockchains
  4. Scaling techniques for smart contracts
  5. Problem Description
  6. Data market for training ML models
  7. Security goals
  8. Solution overview
  9. Preliminaries
  10. Cryptographic primitives
  11. Federated learning with model perturbation
  12. Blockchain and smart contract
  13. OmniLytics+ Data Market
  14. System overview
  15. Validation predicate
  16. ...and 9 more sections

Figures (7)

  • Figure 1: System overview of the proposed OmniLytics+ data market. OmniLytics+ consists of an on-chain smart contract ModelTrade and an off-chain server cluster. The interactions between different parties in a data trading session are illustrated and numbered by their orders of occurrence.
  • Figure 2: State transition of the smart contract ModelTrade. The eight states of ModelTrade are represented by ovals. State transitions are triggered by either applying a method (in a solid box), or occurrence of an event (in a dashed box).
  • Figure 3: MSE of raining MLP on UBMD.
  • Figure 4: Loss of training ResNet20 on CIFAR 10.
  • Figure 5: Gas cost for multiple servers.
  • ...and 2 more figures