Kwek, Shee Hao (2011) Neural Network Activation Function For Writer Indentification : A Comparison. Project Report. UTeM, Melaka, Malaysia. (Submitted)
![[img]](http://digitalcollection.utem.edu.my/style/images/fileicons/application_pdf.png) |
PDF (24 Pages)
Neural_Network_Activation_Function_For_Writer_Indentification__A_Comparison_-_24_pages.pdf - Submitted Version Download (618kB) |
|
PDF (Full Text)
Neural_Network_Activation_Function_For_Writer_Indentification__A_Comparison.pdf - Submitted Version Restricted to Registered users only Download (3MB) |
Abstract
riter identification is use a lot in forensic and biometric field. This proves that entification has a wide variety of application, such as security, financial, forensic rs. The purpose of this research is to compare activation functions for neural in order to develop better performance of identification in writer identification. lt obtain from the testing will be analyzed to determine which has better nce on writer identification. The two activation functions are sinusoidal n function and optimized sigmoid function. The activation function is use to its weighted input signal and applying and output for better classification. The ows that sinusoidal activation function has better performance compare to the d sigmoid function with 3.28798% accuracy higher than 1.52761% accuracy.
| Item Type: | Final Year Project (Project Report) |
|---|---|
| Uncontrolled Keywords: | Neural networks (Computer science), Pattern recognition systems, Writing -- Identification -- Data processing |
| Subjects: | Q Science > Q Science (General) Q Science > QA Mathematics > QA76 Computer software |
| Divisions: | Library > Final Year Project > FTMK |
| Depositing User: | Users 136 not found. |
| Date Deposited: | 15 Oct 2012 03:32 |
| Last Modified: | 28 May 2015 03:41 |
| URI: | http://digitalcollection.utem.edu.my/id/eprint/6296 |
Actions (login required)
 |
View Item |
Altmetric
Altmetric