Uceni s regularizacnimi sitemi

In this work we study and develop learning algorithms for networks based on regularization theory. In particular, we focus on learning possibilities for a family of regularization networks and radial basis function networks (RBF networks). The framework above the basic algorithm derived from theory...

Full description

Bibliographic Details
Main Author: Kudová, Petra
Other Authors: Neruda, Roman
Format: Doctoral Thesis
Language:English
Published: 2007
Online Access:http://www.nusl.cz/ntk/nusl-271005
id ndltd-nusl.cz-oai-invenio.nusl.cz-271005
record_format oai_dc
spelling ndltd-nusl.cz-oai-invenio.nusl.cz-2710052021-02-25T05:16:13Z Uceni s regularizacnimi sitemi Learning with Regularization Networks Kudová, Petra Neruda, Roman Andrejková, Gabriela Hlaváčková-Schindler, Kateřina In this work we study and develop learning algorithms for networks based on regularization theory. In particular, we focus on learning possibilities for a family of regularization networks and radial basis function networks (RBF networks). The framework above the basic algorithm derived from theory is designed. It includes an estimation of a regularization parameter and a kernel function by minimization of cross-validation error. Two composite types of kernel functions are proposed - a sum kernel and a product kernel - in order to deal with heterogenous or large data. Three learning approaches for the RBF networks - the gradient learning, three-step learning, and genetic learning - are discussed. Based on the se, two hybrid approaches are proposed - the four-step learning and the hybrid genetic learning. All learning algorithms for the regularization networks and the RBF networks are studied experimentally and thoroughly compared. We claim that the regularization networks and the RBF networks are comparable in terms of generalization error, but they differ with respect to their model complexity. The regularization network approach usually leads to solutions with higher number of base units, thus, the RBF networks can be used as a 'cheaper' alternative in terms of model size and learning time. 2007 info:eu-repo/semantics/doctoralThesis http://www.nusl.cz/ntk/nusl-271005 eng info:eu-repo/semantics/restrictedAccess
collection NDLTD
language English
format Doctoral Thesis
sources NDLTD
description In this work we study and develop learning algorithms for networks based on regularization theory. In particular, we focus on learning possibilities for a family of regularization networks and radial basis function networks (RBF networks). The framework above the basic algorithm derived from theory is designed. It includes an estimation of a regularization parameter and a kernel function by minimization of cross-validation error. Two composite types of kernel functions are proposed - a sum kernel and a product kernel - in order to deal with heterogenous or large data. Three learning approaches for the RBF networks - the gradient learning, three-step learning, and genetic learning - are discussed. Based on the se, two hybrid approaches are proposed - the four-step learning and the hybrid genetic learning. All learning algorithms for the regularization networks and the RBF networks are studied experimentally and thoroughly compared. We claim that the regularization networks and the RBF networks are comparable in terms of generalization error, but they differ with respect to their model complexity. The regularization network approach usually leads to solutions with higher number of base units, thus, the RBF networks can be used as a 'cheaper' alternative in terms of model size and learning time.
author2 Neruda, Roman
author_facet Neruda, Roman
Kudová, Petra
author Kudová, Petra
spellingShingle Kudová, Petra
Uceni s regularizacnimi sitemi
author_sort Kudová, Petra
title Uceni s regularizacnimi sitemi
title_short Uceni s regularizacnimi sitemi
title_full Uceni s regularizacnimi sitemi
title_fullStr Uceni s regularizacnimi sitemi
title_full_unstemmed Uceni s regularizacnimi sitemi
title_sort uceni s regularizacnimi sitemi
publishDate 2007
url http://www.nusl.cz/ntk/nusl-271005
work_keys_str_mv AT kudovapetra ucenisregularizacnimisitemi
AT kudovapetra learningwithregularizationnetworks
_version_ 1719378589084811264