Deep learning sample labeling method based on online education big data

Abstract

The invention relates to a deep learning sample labeling method based on online education big data. M pieces of online education data to be labelled are inputted, N pieces of data are displayed, and N pieces of displayed data are labelled; when M is greater than N* number of impressions, Randomly select N data from the remaining M‑N* number of impressions, when it is less than N, select all the remaining data; when M is less than N* number of impressions, randomly select N data from M again, and set the number of impressions to 1; After the sum of all data being labeled is greater than the J multiple of M, the number of times the data is labeled is greater than K as the valid data conforming to the classification; the classification library of different types of data is obtained. The invention is based on the deep learning sample labeling method based on online education big data, which improves the identification of online education data, improves user satisfaction, improves user experience, avoids personal subjectivity, releases the dullness of single repetitive work, and also greatly improves user satisfaction. The big one is convenient for all users who mark it.

Description

technical field [0001] The invention relates to an Internet online education system, in particular to a deep learning sample labeling method based on online education big data. Background technique [0002] The main methods for labeling online education documents (videos, audios, pictures, etc.) are as follows: The first is to screen the documents by a single person to achieve labeling, which is a purely manual labor, and the subjectivity of a single person is too strong, resulting in labeling Inaccurate; the second, based on automated annotation, is still in the research stage, and it cannot be applied to image annotation, not to mention video and audio. [0003] The disadvantage of the prior art is that if thousands of files are marked, it is necessary to repeatedly watch audio, video or pictures for screening, which consumes a lot of manpower and material resources, but is very subjective. As a result, the labeled samples are not accurate enough, and finally the output o...

AI Technical Summary

Problems solved by technology

[0002] The marking methods for online education files (video, audio, pictures, etc.) mainly include: first, it is a pure manual labor to realize the marking by screening the files by a single person, and the subjectivity of a single person is too strong, resulting in the marking Inaccurate; the second type, based on automated annotation, is still in the research stage, and cannot be applied to image annotation, let alone video and audio

[0003] The disadvantage of the existing technology is that if thousands of files are marked, it is necessary to repeatedly watch audio, video or pictures for screening, which consumes a lot of manpower and material resources, but is very subjective

As a result, the labeled samples are not accurate enough, and finally the output of deep learning is inaccurate

Images