ANALYSIS OF PERFORMANCE OF BACKPROPAGATION ANN WITH DIFFERENT TRAINING PARAMETERS. IN: NEURAL NETWORKS IN TRANSPORT APPLICATIONS

Three inputs to a typical backpropagation based artificial neural network (ANN) modelling procedure are the number of hidden units, the learning rate (LR), and the momentum constant (MC). These three inputs have a profound effect on the ANN training as well as the resulting behavior of a trained network. This paper follows research done for the purpose of modeling trip generation using regression analysis and ANNs. The paper first presents a brief introduction to the problem of trip generation, and then explains the database used for modeling. The results of backpropagation modeling are also presented, followed by conclusions and recommendations.

• Availability:
  • Find a library where document is available. Order URL: http://worldcat.org/isbn/184014808X
• Corporate Authors:

  Ashgate Publishing Company

  110 Cherry Street, Suite 3-1
  Burlington, VT  United States  05401-3818
• Authors:
  • Faghri, A
  • Sandeep, A
• Publication Date: 1998

Language

• English

Media Info

• Features: Figures; References; Tables;
• Pagination: p. 57-84
• Serial:
  • Atmospheric Environment
  • Publisher: Elsevier
  • ISSN: 1352-2310
  • Serial URL: http://www.sciencedirect.com/science/journal/13522310

Subject/Index Terms

• TRT Terms: Artificial intelligence; Backpropagation; Model atmosphere; Neural networks; Regression analysis; Training; Trip generation
• Subject Areas: Data and Information Technology; Education and Training; Highways; Planning and Forecasting;

Filing Info

• Accession Number: 00796279
• Record Type: Publication
• ISBN: 184014808X
• Files: TRIS
• Created Date: Jul 25 2000 12:00AM