Personalized federal learning method, electronic equipment and computer readable storage medium

The invention discloses a personalized federal learning method, electronic equipment and a computer readable storage medium, and the method comprises the steps: dividing a local data model structure into two parts: a first part is a representation layer, the first part is sent to a server side for parameter aggregation, and a shared representation layer with very high generalization is learned; the second part is a personalization layer, and the part mainly learns local data characteristics of each client. According to the first part, generalization of a data model is guaranteed, when a new client is added, a good effect can be achieved by combining local data for fine tuning based on existing sharing representation, and training cost is low; and the second part ensures the personalization of the client, so that the model can fit more local data characteristics, and a better effect is obtained locally. According to the combination of the two parts, only the first part participates in federal training, the par