Method for determining function for feature extraction and feature engineering system

By sampling different samples multiple times to generate features and determine the objective function, the data volume and resource overhead problems caused by sample reuse in feature engineering are solved, and the efficiency of sampling and feature generation is improved.

CN119719729APending Publication Date: 2025-03-28HUAWEI TECH CO LTD
0 Cites 0 Cited by

Patent Information

Application Number
CN202311294426.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2023-09-28
Publication Date
2025-03-28

AI Technical Summary

Technical Problem

In feature engineering, multiple samples obtained by preliminary sampling are reused in the subsequent feature generation process, resulting in a huge amount of feature data, resulting in high computing power overhead and storage resource overhead.

Method used

The objective function is determined by sampling different samples multiple times, which indicates the operational logic of the sample to generate high-quality features. This method avoids sample reuse, expands the sampling range, and reduces the amount of feature data and computing resource requirements.

Benefits of technology

The sampling efficiency of feature engineering is improved, the amount of feature data, computing power overhead and storage resource overhead are reduced, and the efficiency of feature generation is enhanced.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention discloses a method for determining a function for feature extraction and a feature engineering system. A plurality of first-order features of the first sample are generated, the first-order features of the first sample are used for determining a first function, and the first function is used for extracting the first-order features; extracting a second sample according to the first function to obtain a plurality of first-order features of the second sample, the first sample being different from the second sample; and generating a plurality of second-order features of the second sample according to the plurality of first-order features of the second sample, the plurality of second-order features of the second sample being used for determining a second function, the second function being used for extracting the second-order features, the second function being used for determining an objective function, the objective function being used for extracting N-order features of the sample, N being greater than or equal to 2, and N being an integer. Different samples are sampled for multiple times to generate the features of the samples, and then the target function is determined. The same sample is prevented from being repeatedly used, and different samples are sampled for multiple times, so that the sampling range of the feature engineering is expanded, and the sampling efficiency of the feature engineering is improved.
Need to check novelty before this filing date? Find Prior Art