A dynamic feature selection method for high-dimensional small sample classification task

By using a dynamic feature selection network, a gating network, and an adaptive temperature adjustment module to generate feature importance scores and sparse masks, and combining this with a multi-task backend network for iterative training, the problems of weak model generalization ability and overfitting in high-dimensional small sample scenarios are solved. This achieves feature selection with high discriminativeness and information completeness, thereby improving classification performance.

CN122153375APending Publication Date: 2026-06-05BEIJING UNIV OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
BEIJING UNIV OF TECH
Filing Date
2026-02-13
Publication Date
2026-06-05

AI Technical Summary

Technical Problem

Traditional feature selection methods cannot adapt to individual differences in high-dimensional, small-sample scenarios, resulting in weak model generalization ability and easy overfitting. Furthermore, they fail to fully explore the reconstructed information of the data and the comparative relationship between samples, thus limiting the discriminative power and robustness of feature subsets.

Method used

A dynamic feature selection network is adopted, including a gating network, an adaptive temperature adjustment module, and a threshold filtering module. By generating feature importance scores, feature selection probability distributions, and sparse masks, and combining them with a multi-task backend network for iterative training, adaptive feature selection and high-quality feature extraction are achieved.

Benefits of technology

It significantly improves the accuracy, robustness, and interpretability of high-dimensional few-sample classification tasks, overcomes the weak generalization ability and overfitting problems of traditional static feature selection strategies, and provides interpretable feature selection results for end-to-end training and inference.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122153375A_ABST
    Figure CN122153375A_ABST
Patent Text Reader

Abstract

The application relates to the field of pattern recognition and machine learning, and discloses a dynamic feature selection method for a high-dimensional small sample classification task, which comprises the following steps: based on a gate network, a first importance score corresponding to the feature dimension of a high-dimensional data training sample is generated; based on the first importance score and an adaptive temperature adjustment module, a feature selection probability distribution is generated and input into a threshold screening module to obtain a first sparse mask; according to the first sparse mask and the high-dimensional data training sample, a first screening feature vector is determined and input into a preset multi-task back-end network to output a joint loss value; based on the joint loss value, the dynamic feature selection network is iteratively trained, a high-dimensional data to-be-tested sample is input into the trained dynamic feature selection network to obtain a second screening feature vector, and a classification result is determined according to the second screening feature vector. The scheme of the application realizes adaptive selection and accurate extraction of features, and improves the accuracy and stability of the high-dimensional small sample classification task.
Need to check novelty before this filing date? Find Prior Art