학술논문
A partition based artificial neural network approach in fingerprint identification for security and compared with support vector machine.
Document Type
Article
Author
Source
Subject
*ARTIFICIAL neural networks
*HUMAN fingerprints
*SUPPORT vector machines
*SCHOOL enrollment
*STATISTICS
*
*
*
*
Language
ISSN
0094-243X
Abstract
The aim of the study was to distinguish fingerprints using state-of-the-art artificial neural networks and support vector machines. Materials and methods: This study compared support vector machines and innovative artificial neural network techniques (N=10). The total sample size was calculated using the g-Power software with an alpha of 0.05, an enrollment rate of 0.1, a confidence interval of 95%, and a pretest power of 80%. Results: The ANN achieves 93% accuracy while the SVM classifier achieves 83% accuracy. In SPSS statistical analysis, the accuracy was found to be significant (p<0.05). Conclusion: The novel ANN classifier provides significantly better fingerprinting performance than the SVM classifier. [ABSTRACT FROM AUTHOR]