RESEARCH ON DEEP NEURAL NETWORK LEARNING BASED ON IMPROVED BP ALGORITHM

HUANG Pei

doi:10.16579/j.issn.1001.9669.2018.04.006

您当前的位置：

首页 >

文章列表页 >

RESEARCH ON DEEP NEURAL NETWORK LEARNING BASED ON IMPROVED BP ALGORITHM

更新时间：2022-09-22

- RESEARCH ON DEEP NEURAL NETWORK LEARNING BASED ON IMPROVED BP ALGORITHM
- Journal of Mechanical Strength Vol. 40, Issue 4, Pages: 796-801(2018)
- 作者机构：
  
  1. 江苏建筑职业技术学院机电工程学院
  2. 中国矿业大学信息与电气工程学院
  3. 徐州市机器人技术实验室
- 作者简介：
- 基金信息：
- DOI：10.16579/j.issn.1001.9669.2018.04.006
  CLC：
扫描看全文
HUANG Pei. RESEARCH ON DEEP NEURAL NETWORK LEARNING BASED ON IMPROVED BP ALGORITHM. [J]. 40(4):796-801(2018)
DOI：

HUANG Pei. RESEARCH ON DEEP NEURAL NETWORK LEARNING BASED ON IMPROVED BP ALGORITHM. [J]. 40(4):796-801(2018) DOI： 10.16579/j.issn.1001.9669.2018.04.006.

摘要

深度学习能够使包含多个处理层的计算模型去学习含有多层次抽象表示的数据。这种学习方式在最先进的语音识别、视觉物体识别、物体检测以及许多其它领域,比如生物基因学和医学等都带来了明显的改善。深度学习能够发现大数据中的复杂结构,而卷积神经网络作为深度学习的重要模型之一在处理语音、图像、视频和文本等方面带来了新的突破。它是利用BP算法来引导机器如何从前一层获取误差来调整本层的参数,从而使这些参数更有利于模型的计算。针对传统BP算法存在的收敛速度慢、常陷入局部极小点的不足,提出了一种快速的BP改进算法。利用改进后的卷积神经网络分别在数据集MNIST、英文字符识别以及医学图像中做实验验证,仿真结果表明了该算法的有效性。

Abstract

Deep learning can make the computing model that contains a number of processing layers to learn the data that contains many levels of abstract representation.This kind of learning way in the most advanced speech recognition,visual object recognition,object detection and many other areas,such as biology,genetics and medicine brought significant improvement.Deep learning can find the complex structure of large data,and the convolution neural network as one of the important models of the depth study in the processing of voice,image,video and text,and other aspects of a new breakthrough.It is the use of BP algorithm to guide the machine how to get the error before the layer to adjust the parameters of this layer,so that these parameters are more conducive to the calculation of the model.In view of the shortcomings of traditional BP algorithm,a fast BP algorithm is proposed,which has the disadvantages of slow convergence speed and often falls into local minimum points.The improved convolutional neural network is used to validate the data set MNIST,English character recognition and medical image.The simulation results show the effectiveness of the proposed algorithm.

关键词

深度学习卷积神经网络改进BP算法

Keywords

Deep learningConvolutional neural networkImproved BP algorithm

references

Views

561

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

THE REVIEW OF MECHANICAL FAULT DIAGNOSIS METHODS BASED ON CONVOLUTIONAL NEURAL NETWORK

RESEARCH ABOUT FAULT DIAGNOSIS OF BEARING BASED ON INSTRINSIC TIME SCALE DECOMPOSITION AND CONVOLUTIONAL NEURAL NETWORK

ROLLING BEARING FAULT DIAGNOSIS BASED ON FUSION CNN AND PSO-SVM

RESEARCH ON GEAR BOX FAULT DIAGNOSIS BASED ON DCNN AND XGBOOST ALGORITHM

APPLICATION OF CONVOLUTIONAL NEURAL NETWORK AND CHAOS THEORY IN FAULT DIAGNOSIS OF ROLLING BEARINGS

Related Author

No data

Related Institution

Campus of Shijiazhuang,Army Engineering University

Energy and Power Engineering Institute,University of Shanghai for Science and Technology

Shanghai Key Laboratory of Multiphase Flow and Heat Transfer in Power Engineering

College of Engineering Science and Technology,Shanghai Ocean University

Shanghai Dian Ji University

⁰