Adaptive polynomial filters with individual learning rates for computationally efficient lung tumor motion prediction

Matous Cejnek, Ivo Bukovsky, Noriyasu Homma, Ondřej Líška

Research output: Chapter in Book/Report/Conference proceedingConference contribution

3 Citations (Scopus)

Abstract

This paper presents a study of higher-order neural units as polynomial adaptive filters with multiple-learning-rate gradient descent for 3-D lung tumor motion prediction. The method is compared with single-learning rate gradient descent approaches with and without learning rate normalization. Experimental analysis is done with linear and quadratic neural unit. The influence of correct selection of adaptation parameters and the dependence of learning time on accuracy were experimentally analyzed. The prediction accuracy is nearly equal to recently published results of batch retraining approaches while the computational efficiency is higher for the introduced approach.

Original languageEnglish
Title of host publication2015 International Workshop on Computational Intelligence for Multimedia Understanding, IWCIM 2015
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781467384575
DOIs
Publication statusPublished - 2015 Dec 3
Externally publishedYes
Event2015 International Workshop on Computational Intelligence for Multimedia Understanding, IWCIM 2015 - Prague, Czech Republic
Duration: 2015 Oct 292015 Oct 30

Publication series

Name2015 International Workshop on Computational Intelligence for Multimedia Understanding, IWCIM 2015

Other

Other2015 International Workshop on Computational Intelligence for Multimedia Understanding, IWCIM 2015
CountryCzech Republic
CityPrague
Period15/10/2915/10/30

Keywords

  • Gradient Descent
  • Linear Neural Unit
  • Prediction
  • Quadratic Neural Unit

ASJC Scopus subject areas

  • Classics
  • Artificial Intelligence

Fingerprint Dive into the research topics of 'Adaptive polynomial filters with individual learning rates for computationally efficient lung tumor motion prediction'. Together they form a unique fingerprint.

Cite this