Title page for etd-0623108-160018


[Back to Results | New Search]

URN etd-0623108-160018
Author Wen-Chin Cheng
Author's Email Address maxwin1387@yahoo.com.tw
Statistics This thesis had been viewed 5589 times. Download 644 times.
Department Electrical Engineering
Year 2007
Semester 2
Degree Master
Type of Document
Language English
Title Study on Least Trimmed Squares Artificial Neural Networks
Date of Defense 2008-06-13
Page Count 54
Keyword
  • Artificial Neural Networks
  • Least Trimmed Squares
  • Abstract In this thesis, we study the least trimmed squares artificial neural networks (LTS-ANNs), which are generalization of the least trimmed squares (LTS) estimators frequently used in robust linear parametric regression problems to nonparametric artificial neural networks (ANNs) used for nonlinear regression problems.
    Two training algorithms are proposed in this thesis. The first algorithm is the incremental gradient descent algorithm. In order to speed up the convergence, the second training algorithm is proposed based on recursive least squares (RLS).
    Three illustrative examples are provided to test the performances of robustness against outliers for the classical ANNs and the LTS-ANNs. Simulation results show that upon proper selection of the trimming constant of the learning machines, LTS-ANNs are quite robust against outliers compared with the classical ANNs.
    Advisory Committee
  • Tsu-Tian Lee - chair
  • Jeng-Yih Juang - co-chair
  • Chang-Hua Lien - co-chair
  • Jyh-Horng Jeng - co-chair
  • Jer-Guang Hsieh - advisor
  • Files
  • etd-0623108-160018.pdf
  • Indicate in-campus at 99 year and off-campus access at 99 year.
    Date of Submission 2008-06-23

    [Back to Results | New Search]


    Browse | Search All Available ETDs

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