Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Data processing method, electronic equipment and computer readable medium

A data processing and data technology, applied in the computer field, can solve the problems of high training sample cost, increased collection difficulty, and overall performance degradation, to ensure the effect of noise elimination and solve the high collection cost.

Active Publication Date: 2020-10-27
BEIJING XINTANG SICHUANG EDUCATIONAL TECH CO LTD
View PDF6 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, this processing method will inevitably lead to a decline in overall performance, and the cost of collecting a large number of training samples will increase, and the difficulty of collection will also increase.

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
  • Data processing method, electronic equipment and computer readable medium
  • Data processing method, electronic equipment and computer readable medium
  • Data processing method, electronic equipment and computer readable medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0027] figure 1 It is a schematic flowchart of the data processing method in Embodiment 1 of the present application. Such as figure 1 As shown, the method includes:

[0028] Step S102: Obtain the first feature data and source identification of the data to be processed.

[0029] In this embodiment, the data to be processed may be any type of data, for example, audio data or image data. For different types of data to be processed, feature extraction is performed on them to obtain first feature data, and the types and extraction methods of the first feature data may be different. Those skilled in the art can extract the required first feature data in an appropriate manner according to requirements, which is not limited in this embodiment. The first feature data can be in vector, matrix or other forms.

[0030] For example, if the data to be processed is speech data, the first feature data may be speech acoustic feature data, such as prosody, frequency spectrum, and sound qu...

Embodiment 2

[0055] In this embodiment, in order to clearly describe the data processing solution provided by the embodiment of the present invention. First, a specific example is used to illustrate the structure of the autoencoder.

[0056] Such as figure 2 As shown, it is a structural block diagram of an autoencoder. The self-encoder includes an encoder, a shared feature layer and a decoder, wherein the structure of the encoder and the decoder are symmetrical about the shared feature layer.

[0057] Wherein, the encoder includes a plurality of first unshared hidden units and a first shared hidden unit. A plurality of first unshared hidden units are arranged in parallel for processing the characteristics of the first feature data from different data sources, so as to eliminate the influence of noise, and obtain the second feature data meeting the set standard.

[0058] The first shared implicit unit is set after a plurality of first unshared hidden units, and is used to map the second...

Embodiment 3

[0175] Figure 4 It is a structural block diagram of the data processing device in Embodiment 3 of the present application. Such as Figure 4 As shown, the data processing device includes: an acquisition module 402, configured to acquire the first feature data and source identification of the data to be processed; a determination module 404, configured to determine the corresponding first non-identical data in the self-encoder according to the source identification. A shared hidden unit, the autoencoder includes a plurality of first non-shared hidden units whose parameters are not shared; a first processing module 406, configured to input the first feature data to the determined first non-shared hidden unit Perform noise elimination in the hidden unit, and output second feature data meeting the set standard; a second processing module 408, configured to input the second feature data into the first shared hidden unit of the autoencoder, Map the second feature data to the set ...

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 invention discloses a data processing method, electronic equipment and a computer readable medium. The data processing method comprises the steps of obtaining first feature data and a source identifier of to-be-processed data; determining a first unshared implicit unit corresponding to the source identifier in an auto-encoder according to the source identifier, wherein the auto-encoder comprises a plurality of first unshared implicit units with unshared parameters; inputting the first feature data into the determined first non-shared implicit unit for noise elimination, and outputting second feature data meeting a set standard; inputting the second feature data into a first shared implicit unit of the auto-encoder, mapping the second feature data to a set feature space through the first shared implicit unit, and outputting mapping data; and inputting the mapping data into a shared feature layer of the auto-encoder, and outputting common feature data extracted through the shared feature layer. The data processing method can eliminate noise data in the data.

Description

technical field [0001] The present application relates to the field of computer technology, and in particular to a data processing method, electronic equipment and a computer-readable medium. Background technique [0002] With the development and progress of science and technology, people pay more and more attention to machine learning, and the development of machine learning is also getting faster and faster. In the process of machine learning, the quality of training samples is a crucial factor affecting the performance of machine learning models. [0003] In some scenarios, the data used as training samples may be heterogeneous data of the same type but from different sources. Heterogeneous data means that the noise data contained in it is different due to different collection devices and / or collection environments, which will lead to performance degradation of the trained machine learning model affected by different noise data. Taking voice data as an example, when rec...

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
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/62G06K9/40G06N3/04G06N3/08
CPCG06N3/088G06V10/30G06N3/045G06F18/214G06N3/0455G06N3/08
Inventor 杨嵩黄健杨非刘子韬黄琰
Owner BEIJING XINTANG SICHUANG EDUCATIONAL TECH CO LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com