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

Dynamic missing value filling method based on detracking auto-encoder

An autoencoder and missing value technology, applied in neural learning methods, neural architectures, biological neural network models, etc., can solve problems such as reduced training accuracy, achieve structural simplicity, improve known information utilization, and improve regression performance effect

Inactive Publication Date: 2019-10-08
DALIAN UNIV OF TECH
View PDF2 Cites 3 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Once the missing rate is too large and the number of complete samples decreases, the training accuracy will be greatly reduced

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
  • Dynamic missing value filling method based on detracking auto-encoder
  • Dynamic missing value filling method based on detracking auto-encoder
  • Dynamic missing value filling method based on detracking auto-encoder

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0043] The specific embodiments of the present invention will be described in detail below in combination with the summary of the invention and the accompanying drawings.

[0044] figure 1 It is a working flow diagram of the present invention. In the figure, the first row A in the incomplete data set 1 ,A 2 ,A 3 ,...,A s Indicates attribute names, and black markers indicate missing values. based on figure 1 It can be seen that the present invention builds the network model TRAE according to the number of attributes of the data set, and then uses the filling scheme MVPT to realize network training and filling of missing values ​​in parallel. Before training, the scheme randomly initializes the network parameters and missing value estimates; during the training process, the entire incomplete data set is input into TRAE as a training set; TRAE updates network parameters and missing value estimates based on the optimization algorithm; the updated missing value estimates are ...

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 dynamic missing value filling method based on a detracking auto-encoder, and belongs to the technical field of data mining. The dynamic missing value filling method comprisestwo parts: a network model part and a filling scheme part. In the network model part, in order to strengthen the dependency relationship of missing values in incomplete samples on existing data, a calculation rule of hidden nodes is designed on the basis of a traditional auto-encoder, and a detracking auto-encoder is constructed. In the filling scheme part, for the data incompleteness, a fillingscheme based on a missing value dynamic processing mechanism is designed, and a missing value is regarded as an unknown variable of a cost function, and an estimation value of the unknown variable isdynamically adjusted based on an optimization algorithm, and accompanying expression is completed when network training is finished. According to the dynamic missing value filling method, the learningability of the network model to the cross correlation among the attributes is enhanced, and all data information in the incomplete data set is fully utilized, and ideal filling precision is achieved.

Description

technical field [0001] The invention belongs to the technical field of data mining, and relates to a dynamic missing value filling method based on detracking autoencoders. Background technique [0002] Data mining refers to the process of extracting potentially effective information from massive data based on various algorithms. In data mining, data quality is a key factor affecting the reliability of results. However, due to mistakes in data acquisition, input, storage, operation and other processes, real data sets often have varying degrees of missing and thus reduce data quality. Filling missing values ​​has become an important task in data mining. As a current hot research topic, neural network provides an effective solution for filling missing values. [0003] The missing value filling method based on neural network (P.J.García-Laencina, J.L.Sancho-Gómez, A.R.Figueiras-Vidal. Pattern classification with missing data: a review. Neural Computing and Applications.2010,19...

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): G06N3/04G06N3/08
CPCG06N3/08G06N3/045
Inventor 张立勇赖晓晨吴霞
Owner DALIAN UNIV OF TECH
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