Model training method, apparatus and device, and readable storage medium

Abstract

The invention provides a model training method. The method includes: taking a preset target model of which the loss function value is smaller than or equal to a first preset threshold value as a to-be-tested model; and obtaining a test result of the to-be-tested model, and when the test result meets a preset test condition, taking the to-be-tested model as a prediction model, or when the prediction result does not meet the test condition, updating an influence factor of the sample data according to the test result. According to the model training method provided by the embodiment of the invention, the influence of the sample data on model training can be automatically controlled, and on one hand, poor model training effect caused by the sample data distribution problem in the training process of the model can be avoided, so that the prediction accuracy of the finally obtained prediction model is low; and on the other hand, compared with a manual data cleaning method in the prior art, the situation of missing detection is avoided, and a large amount of labor cost and time are saved.

Description

technical field [0001] The present application relates to the technical field of machine learning, and more specifically, relates to a model training method, device, equipment and readable storage medium. Background technique [0002] The sample data used for model training is often mixed with dirty data or default data, and there may be imbalanced sample data. Existing model training methods use sample data directly for machine learning tasks, which often leads to inaccurate model training results, so that in the actual prediction process of the trained prediction model, the prediction results have a large deviation from the actual one. Contents of the invention [0003] In view of this, the present application provides a model training method, device, equipment and readable storage medium, as follows: [0004] A method for training a model, comprising: [0005] A preset target model whose loss function value is less than or equal to a first preset threshold is used as ...

AI Technical Summary

Problems solved by technology

[0002] The sample data used for model training is often mixed with dirty data or default data, and there may be imbalanced sample data

Existing model training methods use sample data directly for machine learning tasks, which often leads to inaccurate model training results, so that in the actual prediction process of the trained prediction model, the prediction results have a large deviation from the actual

Images