Energy Function Based Evaluation Method of the GeneralizationAbility of Interpolated-based Surrogate Model
DOI:
CSTR:
Author:
Affiliation:

Clc Number:

Fund Project:

  • Article
  • |
  • Figures
  • |
  • Metrics
  • |
  • Reference
  • |
  • Related
  • |
  • Cited by
  • |
  • Materials
  • |
  • Comments
    Abstract:

    The characteristics of interpolated-based surrogate model were analyzed. The oscillation characteristics of interpolated-based surrogate model, an important factor that affect the generalization capacity was put forward. Analysis shows that: when the oscillation of surrogate model is violent, its generalization ability is weak. An evaluation method based on energy function was put forward to evaluate the generalization ability of surrogate model. By calculating the surrogate model’s first and second derivatives, Energy Function Method (EFM) can obtain the oscillation characteristics of surrogate model. The higher the Energy Function, the more serious the oscillation of surrogate model. Thus its generalization ability is worse. Test functions were made, and the results show that the Energy Function Method is efficient in evaluating the generalization ability of the surrogate model.

    Reference
    Related
    Cited by
Get Citation
Share
Article Metrics
  • Abstract:
  • PDF:
  • HTML:
  • Cited by:
History
  • Received:March 28,2011
  • Revised:
  • Adopted:
  • Online: August 28,2012
  • Published:
Article QR Code