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

Model design method based on depth nearest class mean value and incremental smell classification method

A mean value model and depth technology, applied in neural learning methods, biological neural network models, computing, etc., can solve problems such as being unsuitable for highly nonlinear classification tasks, and achieve the effect of wide application range and small calculation amount.

Pending Publication Date: 2019-08-23
GUANGDONG UNIV OF TECH +1
View PDF0 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, it is a linear classifier and is not suitable for highly nonlinear classification tasks (such as odor classification)

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
  • Model design method based on depth nearest class mean value and incremental smell classification method
  • Model design method based on depth nearest class mean value and incremental smell classification method
  • Model design method based on depth nearest class mean value and incremental smell classification method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0063] Such as image 3 As shown, Embodiment 1 of the present invention provides a method for designing a deep nearest class mean model, and the method includes the following steps:

[0064] S101. Design a deep neural network model; wherein, the deep neural network model includes: an input layer, at least one hidden layer, a fully connected layer, and a Softmax functional layer;

[0065] S102. Replace the fully connected layer in the deep neural network model with an NCM layer to obtain a deep nearest class mean model.

[0066] Preferably, after the step of replacing the fully connected layer in the deep neural network model with an NCM layer to obtain the depth nearest class mean model, the method also includes the following model training steps:

[0067] Select training data and test data of the nearest class mean model of depth; Wherein, the training data includes initial training phase data and update training phase data, and the test data includes initial testing phase d...

Embodiment 2

[0078] Such as Figure 4 As shown, Embodiment 2 of the present invention provides an incremental odor classification method based on the depth nearest class mean model, which includes all the contents of the design method of a depth nearest class mean model described in Embodiment 1, and also includes The following steps:

[0079] S201. Collect odor data;

[0080] S202. Sampling the collected odor data through the sensor array to obtain sensor array signals;

[0081] S203. Input the odor data into the pre-trained deep closest mean class model, and perform feature classification by the Softmax function layer in the deep closest mean class model.

[0082] The embodiment of the present invention provides an incremental odor classification method based on the deep nearest class mean model, which combines the advantages of the deep neural network and the nearest class mean classifier, and eliminates their respective limitations, and can seamlessly accommodate new odor categories...

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 model design method based on a depth nearest class mean value and an incremental smell classification method. The method comprises the following steps: designing a depth neural network model, wherein the deep neural network model comprises an input layer, at least one hidden layer, a full connection layer and a Softmax function layer; and then replacing a full connectionlayer in the deep neural network model with an NCM layer to obtain a depth nearest class mean value model. According to the model design method based on the depth nearest class mean value and the incremental smell classification method provided by the embodiment of the invention, the advantages of the depth neural network and the nearest class mean value classifier are combined, respective limitations of the depth neural network and the nearest class mean value classifier are eliminated, a new smell class can be seamlessly accommodated, and retraining is not needed. The method is small in calculation amount and wide in application range.

Description

technical field [0001] The embodiments of the present invention relate to the technical field of gas detection, and in particular to a model design method based on the depth nearest class mean and an incremental odor classification method. Background technique [0002] Odor recognition by electronic noses (E-nose) plays an important role in many applications, such as detection and diagnosis in medicine, finding drugs and explosives, quality control in food production chains and locating gas energy sources. Some odor-based classification algorithms (such as Chinese herbal medicine identification, drug diagnosis, and dangerous goods search) need to update their knowledge over time as they face constantly updated odors. These tasks are viewed as incremental odor classification, where the number of samples and categories in the training dataset is gradually increased. To handle incremental odor classification tasks, non-incremental algorithms have to be retrained from scratch, ...

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): G06K9/62G06N3/04G06N3/08G01N33/00
CPCG06N3/08G01N33/0001G06N3/045G06F18/2414G06F18/24147
Inventor 程昱詹灿坚何家峰骆德汉
Owner GUANGDONG 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