Title page for etd-0618109-174414


[Back to Results | New Search]

URN etd-0618109-174414
Author Shih-hui Liao
Author's Email Address liao622@gmail.com
Statistics This thesis had been viewed 5565 times. Download 1710 times.
Department Electrical Engineering
Year 2008
Semester 2
Degree Master
Type of Document
Language English
Title Study on Additive Generalized Radial Basis Function Networks
Date of Defense 2009-05-29
Page Count 69
Keyword
  • Generalized Radial Basis Function Network
  • Additive Model
  • Additive Generalized Radial Basis Function Network
  • Abstract In this thesis, we propose a new class of learning models, namely the additive generalized radial basis function networks (AGRBFNs), for general nonlinear regression problems. This class of learning machines combines the generalized radial basis function networks (GRBFNs) commonly used in general machine learning problems and the additive models (AMs) frequently encountered in semiparametric regression problems. In statistical regression theory, AM is a good compromise between the linear model and the nonparametric model. In order for more general network structure hoping to address more general data sets, the AMs are embedded in the output layer of the GRBFNs to form the AGRBFNs. Simple weights updating rules based on incremental gradient descent will be derived. Several illustrative examples are provided to compare the performances for the classical GRBFNs and the proposed AGRBFNs. Simulation results show that upon proper selection of the hidden nodes and the bandwidth of the kernel smoother used in additive output layer, AGRBFNs can give better fits than the classical GRBFNs. Furthermore, for the given learning problem, AGRBFNs usually need fewer hidden nodes than those of GRBFNs for the same level of accuracy.
    Advisory Committee
  • Ker-Wei Yu - chair
  • Yih-Lon Lin - co-chair
  • Shiang-Hwua Yu - advisor
  • Jer-Guang Hsieh - advisor
  • Files
  • etd-0618109-174414.pdf
  • indicate accessible in a year
    Date of Submission 2009-06-18

    [Back to Results | New Search]


    Browse | Search All Available ETDs

    If you have more questions or technical problems, please contact eThesys