Node selection and aggregation optimization system and method for federated learning under micro-service architecture

Abstract

The invention discloses a node selection and aggregation optimization system and method for federated learning under a micro-service architecture, and the system comprises a node division module which is used for dividing nodes into data nodes and model nodes according to the specific privacy protection demands; the participation model aggregation node selection module is used for selecting model nodes participating in model aggregation and data nodes for verifying model parameters by the center party according to the node credibility; the node creditworthiness is generated based on the quality of transmission data or a model of the sensing node and the historical creditworthiness of the sensing node; and the credit chain and transaction chain construction module is used for establishing two block chains, namely a credit chain and a transaction chain, and respectively recording the credibility of the nodes and the types of exchange data or models between the federal learning center party and the nodes in each round. According to the method, a specific quality evaluation and model aggregation method is constructed for different privacy requirements of nodes, and the advantages of federated learning on node data privacy protection and non-tampering and traceability of a block chain are brought into full play.

Description

technical field [0001] The invention belongs to the technical field of network privacy and security, and in particular relates to a node selection and aggregation optimization system and method for federated learning under a microservice architecture. Background technique [0002] The microservice architecture is extremely scalable, and code can be easily added, deleted, or modified; technicians can use different stacks for different components; components can be scaled independently of each other, reducing the waste of having to scale the entire application and the characteristics of the cost paid. However, the current microservice architecture is generally unstable. [0003] Crowd intelligence is based on users and their smart terminals (such as mobile phones, wearable devices, smart cars) as the carrier (node), based on the participation of a large number of nodes, the large-scale data collected is uploaded to the task processing platform, and the data collection A micr...

AI Technical Summary

Problems solved by technology

[0005] The present invention aims at the ubiquitous instability of the current microservice architecture, and the existing method of using federated learning to solve the problem of network privacy security in crowd sensing without considering the different requirements of different participating nodes for their own privacy, and proposes a Node selection and aggregation optimization (OAFL) system and method for federated learning under microservice architecture

Images