Paper Title:
A Clustering Method for Pruning Fully Connected Neural Network
  Abstract

This paper focuses mainly on a clustering method for pruning Fully Connected Backpropagation Neural Network (FCBP). The initial neural network is fully connected, after training with sample data, a clustering method is employed to cluster weights between input to hidden layer and from hidden to output layer, and connections that are relatively unnecessary are deleted, thus the initial network becomes a PCBP (Partially Connected Backpropagation) Neural Network. PCBP can be used in prediction or data mining more efficiently than FCBP. At the end of this paper, An experiment is conducted to illustrate the effects of PCBP using the submersible pump repair data set.

  Info
Periodical
Advanced Materials Research (Volumes 204-210)
Edited by
Helen Zhang, Gang Shen and David Jin
Pages
600-603
DOI
10.4028/www.scientific.net/AMR.204-210.600
Citation
G. Li, X. S. Qian, C. M. Ye, L. Zhao, "A Clustering Method for Pruning Fully Connected Neural Network", Advanced Materials Research, Vols. 204-210, pp. 600-603, 2011
Online since
February 2011
Export
Price
$32.00
Share

In order to see related information, you need to Login.

In order to see related information, you need to Login.

Authors: Zhang Qi Hu, Wei Rong Lv, Yuan Li, Chao Fei Wang
Chapter 7: Building Materials
Abstract:In the structure design, the definition of the joints had a great impact on the structure internal force distribution, thus affecting the...
1712
Authors: Hai Rong Xu, Lu Hui Yang, Qiang Huang, Jian Wang
Chapter 15: Software and Computer in Research and Engineering Solutions
Abstract:The Aircraft scheduling problem is researched in this paper, the equilibrium model of aircraft scheduling problem is proposed and a sorting...
2488
Authors: Feng Liu
Chapter 23: Data Mining, Information Technologies and Computer Applications in Industry
Abstract:In the research area of network security, we often need to analyze the Internet traffic in real time. But the Internet traffic is usually...
2506