Machine learning model compression method and system based on federated learning and mean iteration

By adopting federated learning and mean iteration methods in machine learning model compression, redundant network structures and parameters are trimmed, and the problems of low efficiency and insufficient performance of model compression in the prior art are solved, and more efficient computing and storage resource utilization is achieved.

CN119831068APending Publication Date: 2025-04-15CHONGQING UNIV
0 Cites 2 Cited by

Patent Information

Application Number
CN202411791402.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-12-06
Publication Date
2025-04-15

AI Technical Summary

Technical Problem

While ensuring model performance, existing machine learning model compression methods cannot effectively reduce computing and storage overhead, increasing network bandwidth burden and data privacy leakage, and poor overall system performance and scalability.

Method used

Using a machine learning model compression method based on federated learning and mean iteration, the basic model is initialized by the server side, the client locally trains and aggregates parameters, and the redundant network structure and parameters are cropped, combined with mean iteration optimization and parameter screening, and the lightweight model is reorganized.

Benefits of technology

It reduces the risk of network bandwidth burden and data privacy leakage, improves the training quality of the global model and the overall system performance, reduces model redundancy and improves computing efficiency.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention discloses a machine learning model compression method and system based on federated learning and mean iteration, and belongs to the field of deep learning optimization, and the compression method comprises the following specific steps: I, a server initializes a group of basic models, and distributes the models to each client node; iI, updating model parameters by each client node by using local data, and aggregating the client parameters into a global model by the server side; according to the method, the network bandwidth burden and the risk of data privacy leakage can be reduced, effective fusion of training results of different clients can be promoted, the training quality of a global model can be improved, the overall performance and expandability of a system can be remarkably improved, meanwhile, model redundancy is reduced, and the calculation efficiency is improved; the method can further reduce the storage and calculation overhead, improve the compression efficiency, effectively maintain or even improve the precision of the compressed model, ensure the effectiveness of the model at different clients, reduce the communication overhead and calculation burden, and improve the overall efficiency of the system.
Need to check novelty before this filing date? Find Prior Art