Eureka AIR delivers breakthrough ideas for toughest innovation challenges, trusted by R&D personnel around the world.

Evolutionary neural network structure search method

A network structure and search method technology, applied in the field of evolutionary neural network structure search, can solve the problems of slow convergence speed of convolutional neural network

Pending Publication Date: 2021-02-09
ANHUI POLYTECHNIC UNIV MECHANICAL & ELECTRICAL COLLEGE
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] The present invention provides an evolutionary neural network structure search method to solve the problem in the above-mentioned background technology that the training results are overfitted due to the huge network structure and parameters, which will lead to slow convergence of the convolutional neural network. technical issues

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
  • Evolutionary neural network structure search method
  • Evolutionary neural network structure search method
  • Evolutionary neural network structure search method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0032] Refer to attached figure 1 , an evolutionary neural network structure search method, is characterized in that comprising the following steps:

[0033] Step1: Basic parameter setting of the particle swarm algorithm formula, including the population size of the particle swarm algorithm, the maximum number of iterations and other parameters of the particle swarm algorithm;

[0034] Step2: Use the neural network structure parameters as the particle components of the particle swarm algorithm algorithm, and initialize each particle randomly;

[0035] Step3: Evaluate each particle and obtain the global optimum, use each particle as the CNN structure, randomly initialize the CNN weight, determine the CNN through training, and use the CNN test error as the particle's fitness function value, pbest and gbest positions;

[0036] Step4: If the fitness function value of gbest is less than the given threshold or reaches the maximum number of iterations, then stop the iteration, other...

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 provides an evolutionary neural network structure search method, which aims at solving the problem that an existing convolutional neural network is widely applied to the field of machinevision such as object detection, image recognition and classification, but the convergence speed of the convolutional neural network is low due to the fact that a training result is overfitted due toa huge network structure and parameters. The advantages of the particle swarm algorithm in the aspects of the search range and the convergence speed are utilized, the search breadth and the convergence speed can be considered at the same time, the structure discovered by the method can effectively solve the problems, and an interpretable and robust neural network method is established.

Description

technical field [0001] The invention mainly relates to the field of network structure search, in particular to an evolutionary neural network structure search method. Background technique [0002] Convolutional neural network is a kind of feed-forward neural network that includes convolution calculation and has a deep structure. Variation classification, so it is also called "translation invariant artificial neural network". [0003] The existing convolutional neural network is widely used in machine vision fields such as object detection, image recognition and classification, but the training results are overfitted due to the huge network structure and parameters, which will lead to slow convergence of the convolutional neural network. The problem. Contents of the invention [0004] The technical problem to be solved by the invention [0005] The present invention provides an evolutionary neural network structure search method to solve the problem in the above-mentione...

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): G06N3/08G06N3/04
CPCG06N3/086G06N3/045
Inventor 万家山
Owner ANHUI POLYTECHNIC UNIV MECHANICAL & ELECTRICAL COLLEGE
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
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