FEDERATED-LEARNING-BASED PERSONAL QUALIFICATION EVALUATION METHOD, APPARATUS AND SYSTEM, AND STORAGE MEDIUM

摘要

A federated-learning-based personal qualification evaluation method, apparatus and system, and a storage medium. The system comprises: a smart terminal (100), an external participant (200), local participants (300) and a central server end (400), wherein the smart terminal (100) performs training on the basis of user behavior data, so as to obtain a first evaluation sub-model; the external participant (200) sends external user data to the central server end, and the central server end (400) performs training on the basis of the external user data, so as to obtain a second evaluation sub-model; and the local participants (300) send gradients of third evaluation sub-models to the central server end (400), and the central server end (400) calculates the weighted average of the obtained gradients to generate an average gradient, and updates model parameters of the third evaluation sub-models on the basis of the average gradient, such that the local participants (300) train a third evaluation model again. The central server end (400) integrates model parameters of the first evaluation sub-model, model parameters of the second evaluation sub-model and model parameters of the third evaluation sub-models to obtain a final global evaluation model.