一种代理辅助的主从式联邦进化特征选择模型构建方法
By using a proxy-assisted master-slave federated evolutionary feature selection model, combined with joint embedded feature selection and variable step-size recursive feature deletion, the problems of dimensionality curse and high computational cost in high-dimensional feature selection are solved, achieving efficient and privacy-secure feature screening that can adapt to the personalized needs of different participants.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- ANHUI NORMAL UNIV
- Filing Date
- 2025-12-08
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies cannot effectively solve problems such as the curse of dimensionality, high computational cost, data privacy protection, and adaptation to the personalized needs of participants in high-dimensional feature selection. In particular, in federated scenarios, feature selection methods suffer from screening bias and excessive redundant computation.
A master-slave federated evolutionary feature selection model with agent assistance is adopted. Through joint embedded feature selection, agent model collaborative management and variable step size recursive feature deletion, it can efficiently filter feature subsets. This includes the construction and updating of agent models for both the demand side and the assisting side, and dynamically adjusting the feature deletion step size to meet the needs of privacy protection and personalization.
It significantly improves classification accuracy, reduces feature dimensionality and computation time, adapts to datasets of different sizes and category distributions, meets privacy protection requirements, and solves the problems of computational cost and search efficiency in high-dimensional feature selection.
Smart Images
Figure CN121682188B_ABST