The invention discloses a socialized learning method for a
smart city. The method comprises the following steps: constructing a layered socialized learning
system; establishing a task evaluation model based on deep
reinforcement learning, and optimizing the task evaluation model by using task states and channel states of all
the Internet of Things devices to obtain a basic decision; the
edge server utilizes federal learning edge aggregation to receive the task evaluation model, and optimizes the task evaluation model on the
edge server according to the basic decision to obtain a high-level decision; the
edge server uses transfer learning to guide a model in
the Internet of Things equipment; and the
cloud server aggregates the received task evaluation model by using federal learning cloud, formulates a municipal decision according to a high-level decision and the task evaluation model on the
cloud server, and guides the task evaluation model on the edge
server by using transfer learning. According to the method, federal learning is utilized to improve cooperation among intelligent agents in the
layers, transfer learning is utilized between the
layers to realize guidance from the upper layer to the lower layer, and the performance of the model is improved.