• Türkçe
    • English
  • English 
    • Türkçe
    • English
  • Login
View Item 
  •   DSpace Home
  • 2- Enstitüler | Institutes
  • Fen Bilimleri Enstitüsü / Graduate School for Science and Engineering
  • Yüksek Lisans / Master's Programs
  • FBE - Tez Koleksiyonu | Bilgisayar Mühendisliği / Computer Engineering
  • View Item
  •   DSpace Home
  • 2- Enstitüler | Institutes
  • Fen Bilimleri Enstitüsü / Graduate School for Science and Engineering
  • Yüksek Lisans / Master's Programs
  • FBE - Tez Koleksiyonu | Bilgisayar Mühendisliği / Computer Engineering
  • View Item
JavaScript is disabled for your browser. Some features of this site may not work without it.

Spline based neural networks

Thumbnail

View/Open

MasterThesis (875.6Kb)

Date

2005

Author

Dalkılıç, Hikmet

Metadata

Show full item record

Citation

Dalkılıç, H., (2005). Spline based neural networks. İstanbul: Işık Üniversitesi Fen Bilimleri Enstitüsü.

Abstract

In this thesis, we applied the Catmull-Rom splines and B-splines to the neural networks models, which are Multi Layer Perceptrons, Elman Networks, and Locally Recurrent Neural Networks, as adaptive activation functions. We derived the learning algorithms for the five new neural network models, which we proposed. This new models are called 2Multi Layer Perceptrons with Adaptive B- Spline Activation Function3, 2Elman Networks with Adaptive Catmull-Rom Spline Activation Function3, 2Elman Networks with Adaptive B- Spline Activation Function3, 2Locally Recurrent Neural Networks with Adaptive Catmull-Rom Spline Activation Function3, 2Locally Recurrent Neural Networks with Adaptive B- Spline Activation Function3. We measure the performance of these networks on the xor problem and compare the performance of them for this problem. To simulate the networks and to compare their performances we developed a web-based neural network simulator written in PHP 4 called SBNN.
 
Bu tez ile, Catmull-Rom spline fonksiyonları ve B-spline fonksiyonları uyarlanabilir aktivasyon fonksiyonları olarak, yapay sinir ağı modelleri olan Çok Katmanlı Ağlara,Elman ağlarına ve Yerel Geri Beslemeli ağlara uygulandı. Bu uygulamalardan oluşturduğumuz 5 yeni yapay sinir ağı modeli için öğrenme algoritmalarının çıkarımları yapıldı. Bu yeni modeller sırasıyla 2Uyarlanabilir Catmull-Rom Spline Aktivasyon Fonksiyonlu Çok Katmanlı Ağlar3, 2Uyarlanabilir B-Spline Aktivasyon Fonksiyonlu Çok Katmanlı Ağlar3 , 2Uyarlanabilir Catmull-Rom Spline Aktivasyon Fonksiyonlu Elman Ağları3, 2Uyarlanabilir B-Spline Aktivasyon Fonksiyonlu Elman Ağları3, 2Uyarlanabilir Catmull-Rom Spline Aktivasyon Fonksiyonlu Yerel Geri Beslemeli ağlar3, ve son olarak 2Uyarlanabilir B- Spline Aktivasyon Fonksiyonlu Yerel Geri Beslemeli ağlar3 diye adlandırılır. Ağların performansı xor problemi kullanılarak ölçüldü ve sonuçları birbirleriyle karşılaştırıldı. Yapay sinir ağlarını oluşturulması ve performanslarının ölçülmesi için SBNN adında PHP 4 programlama dilin ile yazılmış web tabanlı bir yapay sinir ağı similatörü geliştirildi.
 

URI

https://hdl.handle.net/11729/892

Collections

  • FBE - Tez Koleksiyonu | Bilgisayar Mühendisliği / Computer Engineering [73]

Related items

Showing items related by title, author, creator and subject.

  • Solving fractional diffusion and fractional diffusion-wave equations by Petrov-Galerkin finite element method 

    Esen, Alaattin; Uçar, Yusuf; Yağmurlu, Nuri Murat; Taşbozan, Orkun (Işık University Press, 2014)
    In the last few years, it has become highly evident that fractional calculus has been widely used in several areas of science. Because of this fact, their numerical solutions also have become urgently important. In this ...
  • Fourth-order accurate method based on half-step cubic spline approximations for the 1d time-dependent quasilinear parabolic partial differential equations 

    Mohanty, Ranjan Kumar; Sharma, Sachin (Işık University Press, 2020)
    In this article, we discuss a fourth-order accurate scheme based on cubic spline approximations for the solution of quasilinear parabolic partial differential equations (PDE). The stability of the scheme is discussed using ...
  • A numerical solution of the modified regularized long wave (mrlw) equation using quartic b-splines 

    Karakoç, Seydi Battal Gazi; Geyikli, Turabi; Başhan, Ali (Işık University Press, 2013-09-18)
    In this paper, a numerical solution of the modified regularized long wave (MRLW) equation is obtained by subdomain finite element method using quartic B-spline functions. Solitary wave motion, interaction of two and three ...



DSpace software copyright © 2002-2015  DuraSpace
Contact Us | Send Feedback
Theme by 
@mire NV
 

 




| Policy | Guide | Contact |

DSpace@Işık

by OpenAIRE
Advanced Search

sherpa/romeo

Browse

All of DSpaceCommunities & CollectionsBy Issue DateAuthorsTitlesSubjectsTypeLanguageDepartmentCategoryPublisherAccess TypeIşık AuthorCitationThis CollectionBy Issue DateAuthorsTitlesSubjectsTypeLanguageDepartmentCategoryPublisherAccess TypeIşık AuthorCitation

My Account

LoginRegister

Statistics

View Google Analytics Statistics

DSpace software copyright © 2002-2015  DuraSpace
Contact Us | Send Feedback
Theme by 
@mire NV
 

 


|| Policy || Guide || Library || Işık University || OAI-PMH ||

Işık University Library, Şile, İstanbul, Turkey
If you find any errors in content please report us

Creative Commons License
Işık University Institutional Repository is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 Unported License..

DSpace@Işık:


DSpace 6.2

tarafından İdeal DSpace hizmetleri çerçevesinde özelleştirilerek kurulmuştur.