STRESS-STRAIN MODELING OF SANDS USING ARTIFICIAL NEURAL NETWORKS. DISCUSSION AND CLOSURE

A discussion of a paper with the aforementioned title by Ellis, Yao, Zhao, and Penumadu, published in this journal (Volume 121, Number 5, May 1995), is presented. Discussers Najjar and Basheer report that training for sequential and regular artificial neural networks are the same; the only difference in this phase is the architecture of the network. The discussers also note that the authors' constant incremental strain value for the entire training and testing phases limit the user from the flexibility of utilizing any desired incremental strain value as part of the input. In addition, there is no justification for the value selected. Finally, Najjar and Basheer request discussion on the selection criteria and methodology to optimize both neural networks. Discussion is followed by closure from Penumadu.

• Availability:
  • Find a library where document is available. Order URL: http://worldcat.org/issn/07339410
• Corporate Authors:

  American Society of Civil Engineers

  345 East 47th Street
  New York, NY  United States  10017-2398
• Authors:
  • Ellis, G W
  • Yao, C
  • Zhao, Rui
  • Penumadu, D
• Discussers:
  • Najjar, Y M
  • Basheer, I A
• Publication Date: 1996-11

Language

• English

Media Info

• Features: Figures; Tables;
• Pagination: p. 949-951
• Serial:
  • Journal of Geotechnical Engineering
  • Volume: 122
  • Issue Number: 11
  • Publisher: American Society of Civil Engineers
  • ISSN: 0733-9410

Subject/Index Terms

• TRT Terms: Deformation curve; Mathematical models; Methodology; Neural networks; Optimization; Sand; Training
• Uncontrolled Terms: Criteria; Models; Selection
• Subject Areas: Education and Training; Geotechnology; Highways; I42: Soil Mechanics;

Filing Info

• Accession Number: 00730017
• Record Type: Publication
• Files: TRIS
• Created Date: Dec 30 1997 12:00AM