Machine learning model obtaining method and obtaining device, apparatus, and storage medium

A machine learning model and acquisition method technology, applied in the field of machine learning, can solve the problems of reducing the accuracy of sample pictures, affecting the performance of machine learning models, and uneven levels of annotators

Active Publication Date: 2018-12-18
BEIJING KINGSOFT CLOUD NETWORK TECH CO LTD +1
View PDF19 Cites 14 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, with this training method, mistakes are inevitable in the process of manually labeling pictures, and because the level of labelers is uneven, it is easy to use pictures...

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Machine learning model obtaining method and obtaining device, apparatus, and storage medium
  • Machine learning model obtaining method and obtaining device, apparatus, and storage medium
  • Machine learning model obtaining method and obtaining device, apparatus, and storage medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0097] Such as figure 1 As shown, the embodiment of the present invention provides a method for labeling pictures, which can be applied to servers, and of course, can also be applied to other types of devices with data processing and computing functions. The server is used as an example for illustration below. Include the following steps:

[0098] S101. Acquire multiple pictures of categories to be labeled, and divide the multiple pictures into several parts.

[0099] In the embodiment of the present invention, the server can acquire multiple pictures. Take the labeled icons as training samples for a machine learning model used for class prediction of live video content as an example. These pictures need to be screenshots of the live video in the live room. Specifically, these pictures can include A screenshot of the live video in a live video room, or multiple screenshots taken at multiple points in time from the live video in a live video room. Correspondingly, since the ...

Embodiment 2

[0143] Such as image 3 As shown, the embodiment of the present invention also provides a picture labeling method, which can be applied to the server, and of course can also be applied to other types of devices with data processing and computing functions. The process may include the following steps:

[0144] S301. Use the pictures whose labeled categories are determined as training samples of the machine learning model to obtain the machine learning model.

[0145] In the embodiment of the present invention, the pictures whose marked categories are determined can be used as training samples of the machine learning model, so as to train the machine learning model and obtain the machine learning model for predicting the picture category. Wherein, the above-mentioned picture whose category is determined to be marked may be a picture after the category is determined by the labeling method of the method embodiment 1, and of course it may also be other pictures marked with a catego...

Embodiment 3

[0161] Such as Figure 4 As shown, the embodiment of the present invention also provides a method for acquiring a machine learning model, which can be applied to a server, and of course, can also be applied to other types of devices with data processing functions. In this embodiment, the process of training the machine learning model and obtaining the machine learning model, such as Figure 4 As shown, specifically, the following steps may be included:

[0162] S401. Acquire a first sample picture labeled with a category.

[0163] The first sample picture is a picture that has been marked with a category, for example, a picture that has been marked with a category by an annotator and the label category of the picture has been determined.

[0164] Specifically, the first sample picture marked with a category can be obtained through the labeling method in Method Embodiment 1.

[0165] S402. Using the first sample picture labeled with a category as a training sample, train a m...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The embodiment of the invention provides a machine learning model obtaining method and obtaining device, apparatus, and storage medium, wherein the method comprises the following steps: obtaining a first sample picture labeled with a category; using the first sample picture labeled with category as a training sample, and obtaining the machine learning model by training; inputting a second sample picture to a machine learning model to obtain a category of the second sample picture predicted by the machine learning model; determining a second sample picture having a correct category predicted bythe machine learning model; obtaining a new machine learning model by using the second sample image with correct classification as the training sample. The obtaining method of the machine learning model provided by the embodiment of the invention can improve the prediction accuracy of the machine learning model, that is, improve the performance of the machine learning model.

Description

technical field [0001] The present invention relates to the technical field of machine learning, in particular to an acquisition method, acquisition device, equipment and storage medium of a machine learning model. Background technique [0002] With the popularity of live video broadcasting, a large amount of vulgar pornography and other undesirable content is produced in the live video content. Therefore, it is necessary to effectively supervise the live video content. At present, each live broadcast platform generally supervises the live broadcast content manually, deploys a supervision team of up to hundreds of people, and uses the team members to inspect the live broadcast room to identify bad live broadcast content. But this regulatory approach is costly and inefficient. With the continuous development of artificial intelligence and machine learning technology, deep learning technology can be used to realize automatic identification of video content by machines. [00...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
IPC IPC(8): G06K9/62
CPCG06F18/217
Inventor 刘世权刘弘也苏驰
Owner BEIJING KINGSOFT CLOUD NETWORK TECH CO LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap