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

Method for classifying multi-variable time sequences based on deep long/short-term memory neural network

A long-short-term memory and neural network technology, applied in the field of multivariate time series classification based on deep long-term and short-term memory neural network, can solve the problems of long sample time, low classification accuracy, large number of sample features, etc. High adaptability and high precision

Active Publication Date: 2018-06-19
SOUTH CHINA UNIV OF TECH
View PDF4 Cites 20 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] For multivariate time series classification tasks, some samples have a large number of features, some samples have a very long time length, and some samples belong to a large number of categories, so the univariate time series classification algorithm cannot be simply applied
However, when the single-layer LSTM model faces the above problems, the classification accuracy is not high.

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
  • Method for classifying multi-variable time sequences based on deep long/short-term memory neural network
  • Method for classifying multi-variable time sequences based on deep long/short-term memory neural network
  • Method for classifying multi-variable time sequences based on deep long/short-term memory neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0032] as attached figure 1 as shown, figure 1 is a schematic diagram of the structure of the deep LSTM classification model proposed by the present invention, figure 1 A 3-layer LSTM is used as an example to show the structure of the model.

[0033] A deep recurrent neural network classification model suitable for classification tasks based on long-term short-term memory neurons, using LSTM as the hidden layer of the recurrent neural network, and forming a huge hidden layer state space by stacking LSTM layers, and combining the last LSTM layer The output results are input into the Softmax classifier for classification.

[0034] The long-short-term memory neuron-based recurrent neural network classification model uses a long-short-term memory neuron module including an input gate, an output gate, a forget gate and a state unit, and does not include peephole connections. The long-short-term memory recurrent neural network used, the calculation formula of the output result is...

Embodiment 2

[0053] In this embodiment, for multivariate time series classification problems, a deep recursive neural network model is constructed by using the hidden layer results of a recurrent neural network such as LSTM, and finally a Softmax classifier is used to process the output results of the recurrent neural network, then the multivariable Classify time series data.

[0054] The present invention has universal applicability to time series data sets in various fields such as medicine, machinery, handwriting recognition, language recognition, etc. Now take the speech recognition data set Australian language as an example, which is derived from the UCI machine learning database [K.Bache and M.Lichman, "UCI machine learning repository," 2013. https: / / archive.ics.uci.edu / ml / datasets.html.], the Australian language dataset has a total of 2565 samples, and the number of variables in each sample is 22, the sample length is between [45,136], and belongs to one of the 95 categories. In or...

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 method for classifying multi-variable time sequences based on a deep long / short-term memory neural network. The method includes the steps of by selecting long / short-term memory neuron structures as hidden layer neuron structures of a recurrent neural network, overlapping long / short-term memory neurons, and designing a deep layer recurrent neural network classification framework, thereby achieving the aim of improving the classification accuracy of multi-variable time sequence data. It is found through experimental comparison that accuracy is higher compared with an existing classification model and universality is achieved on time sequence data set classification tasks in multiple fields.

Description

technical field [0001] The invention relates to the technical field of time series data mining, in particular to a method for classifying multivariate time series based on a deep long-short-term memory neural network. Background technique [0002] Multivariate time series data is an important type of time series data, which is used in many fields including medical care, finance, industrial manufacturing, voice, video, etc. The classification of multivariate time series is a basic problem in time series data mining . Compared with the traditional classification model, the multivariate time series classification model mainly has two key points, modeling the correlation between multiple variables and modeling the time series in the data. [0003] At present, the methods for classifying multivariate time series data can be summarized into four categories. One is to use sliding time windows to reconstruct the sample space, introduce time series characteristics into the samples, ...

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): G06F17/30G06K9/62
CPCG06F16/35G06F18/214
Inventor 马千里秦州
Owner SOUTH CHINA 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