一种模型训练的方法、装置、电子设备及存储介质
By collaboratively training edge nodes and central servers, and updating the risk control model using training samples from different business providers, the problem of insufficient sample quantity and categories is solved, achieving higher risk control accuracy and data security.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- ALIPAY (HANGZHOU) INFORMATION TECH CO LTD
- Filing Date
- 2022-12-30
- Publication Date
- 2026-07-17
AI Technical Summary
In existing technologies, business providers can only train risk control models using their own training samples, resulting in a small number of samples and a limited number of categories. This makes it impossible to effectively control user risk, affecting the accuracy of the risk control model and data security.
Training samples from different business providers are obtained through edge nodes to train local risk control models and generate gradient information. This information is then uploaded to the central server for model parameter updates. After aggregating the gradient information, the risk control model on the central server is updated. The updated parameters are then returned to the edge nodes for local risk control model updates, ensuring the consistency of risk control models across nodes.
This improves the accuracy of the risk control model, enabling it to learn the characteristics of training samples from all business providers, thereby enhancing the accuracy of user risk control and data security.
Smart Images
Figure CN116011815B_ABST