Content classification method and device and readable storage medium

Abstract

The invention discloses a content classification method and device and a readable storage medium, and relates to the field of artificial intelligence. The method comprises the steps of obtaining target content; performing classification identification on the target content through a content classification model to obtain a classification result of the target content; and making uncertainty analysis on the target content and the classification result to obtain uncertainty of the classification result, wherein the uncertainty is used for indicating credibility of the classification result, and the uncertainty analysis comprises at least one of inter-class uncertainty analysis and intra-class uncertainty analysis. Uncertainty analysis is carried out on the target content and the classification result so that the inter-class uncertainty and the intra-class uncertainty in the content classification process are combined with the inter-class uncertainty and the intra-class uncertainty to carry out credibility auxiliary analysis on the content classification result of the content classification model, the analysis dimension of the classification result of the content classification model is increased, and the classification accuracy of the content classification model is improved.

Description

Technical field [0001] The embodiments of the present application relate to the field of artificial intelligence, and in particular to a content classification method, device, and readable storage medium. Background technique [0002] Artificial Intelligence (AI) is a theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to obtain the best results. Among them, the neural network model is an implementation of artificial intelligence. [0003] In related technologies, the neural network model can be supervised and trained through labeled samples. Take the image classification model as an example. After inputting the image labeled with the category label into the image classification model for classification, the image is classified according to the label and classification results. The model undergoes su...

AI Technical Summary

Problems solved by technology

[0004] However, since the labels annotated by samples are usually manually annotated according to the results of eye recognition, for samples with small inter-class differences or large intra-class differences, the labeling of the lab

Images