Training data equalization processing method and device, equipment, medium and product
By generating a set of associated data for bank users, obtaining other users and objects of the same category, and generating new associated data to balance the training samples, the problem of unbalanced distribution of user features is solved, the learning and recognition capabilities of the model are improved, overfitting is reduced, and classification accuracy is enhanced.
CN121834736APending Publication Date: 2026-04-10CHINA CONSTRUCTION BANK +1
0 Cites 0 Cited by
Patent Information
- Application Number
- CN202511979959.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-25
- Publication Date
- 2026-04-10
AI Technical Summary
Technical Problem
The uneven distribution of bank user characteristics leads to overfitting of the majority user characteristics, which account for a large proportion, while the model is not good at learning and recognizing the minority user characteristics, which account for a small proportion.
Method used
By generating a set of related data, other users and objects of the same category as the target user are obtained, new related data is generated to balance the training samples, and the related data set is used to train the association relationship classification model.
Benefits of technology
It improves the learning and recognition capabilities of minority user characteristics, reduces model overfitting, and improves model classification accuracy.
✦ Generated by Eureka AI based on patent content.
Abstract
The invention discloses a training data balance processing method and device, equipment, a medium and a product. The invention relates to the technical field of data processing. The method comprises the following steps: when a user in a user set is associated with an object in an object set, generating associated data according to the associated user and object, and adding the associated data into an associated data set; acquiring other users of the same category as the target user corresponding to the to-be-supplemented data of the associated data set; obtaining a target object associated with the target user in the associated data set; according to each target object, determining candidate objects of the target user in other objects associated with the other users in the associated data set; and generating associated data according to the target user and each candidate object, and adding the associated data into the associated data set. The embodiment of the invention can balance the training data of the model.
Need to check novelty before this filing date? Find Prior Art