Evaluation model selection method and device, computer equipment and storage medium

Abstract

The invention discloses an evaluation model selection method and device, computer equipment and a storage medium. The evaluation model selection method comprises the steps: inputting training samplesinto an evaluation model in batches for training; according to a preset rule, selecting an evaluation model needing to be subjected to a first test from the evaluation models trained in batches, and marking the evaluation model as a to-be-evaluated model; respectively inputting a first preset number of first test samples into each to-be-evaluated model, carrying out the first test, obtaining a first test result, and calculating first accuracy of each to-be-evaluated model; after training of the evaluation models is completed, selecting a specified number of preferred evaluation models, carrying out a second test, and calculating a second accuracy rate of each preferred evaluation model; and selecting an optimal evaluation model according to the second accuracy. According to the invention,the optimal evaluation model can be quickly selected from a plurality of training evaluation models.

Description

technical field [0001] The invention relates to the field of machine learning, in particular to an evaluation model selection method, device, computer equipment and storage medium. Background technique [0002] Machine learning models are used in more and more scenarios, such as face recognition, data classification, automatic scoring, etc. Taking data classification as an example, different classification models process the same data, and the accuracy of the classification results is different. Therefore, choosing an appropriate machine learning model can effectively improve the accuracy of data classification. However, for some massive data, a large number of models will be generated in the process of evaluating the model, and these evaluation models need to be tested when determining the optimal evaluation model. However, according to the existing test method, if a large batch of test samples is used for testing, the evaluation effect is good, but it takes a lot of time...

AI Technical Summary

Problems solved by technology

However, according to the existing test method, if a large batch of test samples is used for testing, the evaluation effect is good, but it takes a lot of time and the efficiency is not high; while using a small batch of test samples for testing, the processing efficiency is higher , but its evaluation effect is poor and cannot objectively reflect the true prediction quality of the evaluation model

Images