Federal modeling method and device, electronic equipment, storage medium and computer program

The invention discloses a federated modeling method and device, electronic equipment, a storage medium and a computer program, and relates to the technical field of data processing, P2P point-to-point and centralized aggregation node modes are adopted, a network topology structure of preliminary federated learning is constructed under an unknown network topology through detection message sending and ad-hoc network modes, and the network topology structure of the preliminary federated learning is constructed under the unknown network topology. A deep learning model with a GNN graph neural network as a core is used for graph structure construction and node representation learning, and a task-driven global optimal network topology path and an execution strategy are formed in combination with specific requirements of federated learning tasks; in the federation modeling process, participating nodes form a multi-region network by automatically discovering neighbor nodes, dynamically elect an optimal aggregation nod