Asymmetric Fuzzy Neural Networks Development and Implement on Cloud Systems
| Autor: | Yu-Yang Ho, 何羽洋 |
|---|---|
| Rok vydání: | 2014 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 102 Fuzzy neural networks (FNN) can be regarded as intelligent algorithm that combined with fuzzy logical systems and neural networks. In this thesis, we proposed asymmetric fuzzy neural networks and implemented it on cloud systems under android web service. In the cloud systems under android web service, users can send parameters to the system programs with Matlab/Java in the proposed systems that they have written on the server systems while their android phone is online. Hence, users don’t need to burden the hardware devices. Besides, communication protocol in the proposed systems used the web service. Consequently, user can modify parameters for controlling the proposed program and obtain results from the proposed systems by android phone arbitrarily. In the asymmetric fuzzy neural networks, we apply support vector regression (SVR) and asymmetric influence function to the FNN. That is, the proposed approach has two procedure; namely, initial procedure and learning procedure. A support vector regression is used to do the initial procedure on FNN. Then the learning procedure on FNN cooperated with asymmetric influence function to enhance the FNN for asymmetric noise. The simulation used some functions and systems to verify the proposed approach has better performance. Besides, we also proposed the parameter update method for the asymmetric fuzzy neural networks. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|