A federated learning method, device, client device and storage medium
By conducting specific training on rare categories with client devices whose recognition accuracy is below the threshold in federated learning, the problem of low recognition accuracy of rare categories is solved, and a higher recognition accuracy is achieved.
CN119313980BActive Publication Date: 2025-03-18湖南工商大学
2 Cites 0 Cited by
Patent Information
- Application Number
- CN202411846219.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-16
- Publication Date
- 2025-03-18
- Estimated Expiration
- 2044-12-16
AI Technical Summary
Technical Problem
Client devices have low accuracy in identifying rare categories when predicting or classifying.
Method used
Through a federated learning-based approach, a model is trained to identify rare categories. The specific steps include: training the model based on the preset training set, marking the category with the recognition accuracy below the threshold is a rare category, generating a new training set containing rare samples, and further training the model until the recognition accuracy reaches the threshold.
Benefits of technology
Improves the accuracy of client devices in recognition of rare categories, ensuring that the model performs in rare categories to meet preset standards.
✦ Generated by Eureka AI based on patent content.
Abstract
The present application discloses a federated learning method, apparatus, client device and storage medium, the method comprising: training a federated learning model based on a first training set, using the trained federated learning model as a first model, obtaining a first test result of the first model on a test set, and obtaining a first recognition accuracy rate corresponding to an abnormal category in the first test result; marking an abnormal category whose first recognition accuracy rate is lower than a threshold as a rare category, and forming a second training set with multiple rare categories and multiple rare samples; inputting the rare samples of the second training set into the first model; training the first model, and when the average loss value is less than a preset loss value, stopping training the first model, and using the trained first model as a second model; and saving the second model when the second recognition accuracy rates corresponding to each rare category are greater than a threshold and the set is an empty set. The present application can improve the recognition accuracy rate of the client device for rare categories.
Need to check novelty before this filing date? Find Prior Art