Initialization and training improvements for conic section function neural networks
2011
0 views
0 downloads
Advisor: Yrd. Doç. Dr. Mutlu Avcı
Abstract (EN)
In this study, Genetic Algorithm is used to determine the center vectors of Conic Section Function Neural Networks on initialization phase and improved cone folding method is proposed for training phase of Conic Section Function Neural Networks. The test performances of the Conic Section Function Neural Network with the proposed training versus Conic Section Function Neural Network with classical training are done on Iris, Lenses, Wine, Ecoli and Haberman datasets of UCI machine learning repository and Two Spiral Problem dataset.
Author
Ceyhun Çelik
How to Cite
Ceyhun Çelik (Master Thesis). Initialization and training improvements for conic section function neural networks, 2011, Çukurova University.
License
Tüm Hakları Saklıdır
This work is shared under the specified license terms.
More theses from Çukurova University
- The effects of collaborative video-blog projects on Turkish EFL students' linguistic and digital literacy skills(2025)
- An investigation of violent and nonviolent adolescent' families in terms in terms of family fuctioning, anger and anger expression(2006)
- Adolescents who have single parents family and full family were compared in respect to their life satisfaction and quality of life(2009)
- Credit risk management in banking sector: An application of variables determining credit risk in Turkish banking sector(2011)
- Investigation of psychological symptom levels in adolescents according to gender and family functions(2013)
- Assessing morphological and genetic diversity among traditional African eggplant landraces and detecting salt tolerance and anther culture performance of selected accessions(2022)