Machine learning model feature screening method and device based on data privacy protection

Abstract

The embodiment of the invention provides a machine learning model feature screening method based on data privacy protection. The machine learning model has a set of model features, and feature data ofthe machine learning model is distributed vertically segmented at a first data owner and at least one second data owner, each data owner having feature data corresponding to a subset of model features of the set of model features; the first data owner has the feature data of the to-be-screened model features, and the second data owner does not have the feature data of the to-be-screened model features. And the first data owner and the at least one second data owner cooperatively use the respective feature data to perform multi-party security calculation so as to train a prediction model of the to-be-screened model features. A variance expansion factor of the to-be-screened model feature is determined based on the prediction difference value of the to-be-screened model feature at the firstdata owner so as to perform model feature screening processing.

Description

technical field [0001] The embodiments of this specification generally relate to the computer field, and more specifically, relate to a method and device for feature screening of a machine learning model based on data privacy protection. Background technique [0002] When a company or enterprise conducts business operations, it usually uses machine learning models for model prediction, for example, to determine business categories, business operation risks, or make business operation decisions. The machine learning model may include, for example, a business risk identification model, a business classification model, a business decision model, and the like. A machine learning model usually uses a large number of model features as model input features. For example, a machine learning model may use up to tens of thousands of model features. The greater the number of model features, the greater the computational load of the machine learning model, which requires more computing ...

AI Technical Summary

Problems solved by technology

However, among the model features determined in this way, there may be model features that have little effect on the model effect of the machine learning model. The introduction of these model features will increase the amount of calculation of the machine learning model, but will not improve the performance of the machine learning model. The prediction accuracy of the learning model, therefore, how to screen the model features of the machine learning model has become an urgent problem to be solved

Images