Federal learning method based on directed acyclic graph (DAG) block chain

The invention provides a federal learning method based on a directed acyclic graph (DAG) block chain, and the method comprises the steps: obtaining a current DAG block chain after receiving a training continuing instruction sent by a management device; determining an initial model parameter based on the local data test set and the model parameter in the node in the current DAG block chain; based on the local data training set, training a preset model with the initial model parameters to obtain an intermediate model; based on the model precision of the intermediate model, compressing model parameters of the intermediate model to obtain current target model parameters; wherein the current target model parameter is used for updating the current DAG block chain to obtain a new DAG block chain; and sending the current target model parameter to the management equipment. The federated learning method based on the directed acyclic graph DAG block chain provided by the invention is used for saving communication overhe