A joint learning method and system based on edge computing

Abstract

The invention discloses a joint learning method and system based on edge computing. The method includes: calculating the amount of training data allocated to each training role according to the total amount of sample data owned by each edge cloud and hardware resource parameters of each training role; According to the amount of training data of each training role and the sample data owned by each edge cloud, a data interaction instruction is generated and issued to the corresponding edge cloud, where the data interaction instruction is used to control the redistribution of the sample data among the edge clouds; The model is trained based on the amount of training data allocated by itself until the model converges, and the converged model based on the amount of training data uploaded by each edge cloud is obtained to form a model set; the present invention balances the hardware resources of each training role Training tasks avoid excessive load pressure on the central cloud or an edge cloud, and solve the load pressure on the central cloud after the data is centralized in the central cloud.

Description

technical field [0001] The invention belongs to the technical field of cloud computing, and more particularly, relates to a joint learning method and system based on edge computing. Background technique [0002] With the rapid development of 5G, Internet of Things and other technologies, the centralized computing services provided by cloud computing can no longer meet the needs of the terminal side. It is necessary to provide a distributed open platform that integrates network, computing, storage, and application core capabilities. To provide services, edge computing came into being. With the rapid development and wide application of artificial intelligence (AI), the demand for intelligent computing in terminals is increasing. How to provide intelligent services nearby is the research focus of AI application in edge computing, especially in intelligent industrial control, areas such as autonomous driving and pattern recognition. At present, the construction method of AI mo...

AI Technical Summary

Problems solved by technology

[0005] 1. Data security: The data of each node is collected to the center, and data security protection and legal management will be important issues;

[0006] 2. Privacy protection: Each data owning node may have sensitive data, or there may be issues such as data protection policies and regulations, which may cause the data to be unable to be aggregated and shared;

[0007] 3. Load balancing: The central cloud receives all edge cloud data and conducts centralized model training and updating. The network bandwidth and computing load are relatively large, especially in the field of video image processing. Aggregating a large amount of data will consume a large amount of network resources.

Images