Privacy protection and verifiable federated learning method based on zero knowledge proof

Abstract

The invention relates to a federated learning method for privacy protection and verifiable privacy based on zero knowledge proof, which belongs to the technical field of artificial intelligence machine learning and comprises the steps of training task release, local training, proof generation, training result submission, training process verification and training parameter aggregation. According to the method, the correctness of the training process is proved to the publisher under the condition that the privacy data of the trainer is not leaked by utilizing a zero-knowledge proving technology in the federation learning process. According to the method, a training algorithm used in federated learning is not limited and required, and the proving of any training process is supported, so that the federated learning has the properties of verifiability and privacy protection, and the safety of the federated learning is improved. Meanwhile, a method for converting a decimal machine learning process into an integer machine learning process is adopted, the complex machine learning process is expressed through a series of simple operation combinations related to addition, subtraction, multiplication and division, and the machine learning process and cryptography are organically connected and combined.

Description

technical field [0001] The invention belongs to the technical field of artificial intelligence machine learning, relates to a federated learning method of privacy protection and verifiable privacy, and in particular relates to a federated learning method of privacy protection and verifiable privacy based on zero-knowledge proof. Background technique [0002] As a promising technology, machine learning has become a research hotspot in the computer field, and its theories and methods have been widely used in engineering applications and scientific fields. [0003] Federated learning is a machine learning technology that uses multiple decentralized edge devices or servers to train local models through local data, iteratively fuses local model parameters, and finally calculates global model parameters. Unlike traditional centralized machine learning techniques, federated learning does not require trainers to share data. Therefore, this technology helps to solve the problems of ...

AI Technical Summary

Problems solved by technology

[0007] The purpose of the present invention is to solve the technical problem of how to perform verifiable and secure multi-party training in an untrusted distributed system environment in federated learning, and creatively propose a privacy protection and verifiable method based on zero-knowledge proof federated learning method

Images