1results about How to "Save communication bandwidth" patented technology

This invention belongs to the field of machine learning and discloses a method for updating model parameters in distributed federated learning. It addresses the problems of accuracy loss and reduced convergence rate caused by discarding the local correlation information implicit in gradient values ​​in existing communication compression schemes used during model parameter updates. In this invention, edge nodes use locally stored training samples to calculate the gradient matrix of the model parameters for the current round using the backpropagation algorithm. Then, a pooling operator is used to compress the gradient matrix to obtain the sparse gradient matrix for the current round, which is then uploaded to the central server. The central server, based on the received sparse gradient matrices of each designated edge node for the current round, updates the model parameters for the current round using a distributed gradient descent algorithm. If the updated model parameters meet the required preset accuracy, they are output as the final model parameters; otherwise, the updated model parameters are redistributed to the edge nodes for the next iteration.