Publication: Optimizing convolutional neural network architecture to detect genuine and forgery signatures by using particle swarm optimization
| datacite.subject.fos | oecd::Engineering and technology::Electrical engineering, Electronic engineering, Information engineering | |
| dc.contributor.author | Korniane @ Kornain, Muhammad Mustaqim | |
| dc.date.accessioned | 2024-02-29T07:48:47Z | |
| dc.date.available | 2024-02-29T07:48:47Z | |
| dc.date.issued | 2022-08-01 | |
| dc.description.abstract | This project presents the approach to design convolutional neural network architecture in detecting genuine and forgery signatures with optimization from particle swarm optimization. The important hyperparameter of the convolutional neural network; the convolutional layer, filter number, and filter size in training, and the challenge to adjust them are discussed. Besides, the effect of varying the static parameter of particle swarm optimization in generating the particle such as the inertia weight, social constant, and cognitive constant are analyzed to optimize the neural network and particle swarm optimization algorithm. The method used is also presented in this project. The results obtain for two input datasets, writer dependent and writer independent are 94.44 and 80.28 respectively. Hence, it proved that CNN model can trace the genuine and forgery signature with the optimization by PSO algorithm. | |
| dc.identifier.uri | https://erepo.usm.my/handle/123456789/18521 | |
| dc.language.iso | en | |
| dc.title | Optimizing convolutional neural network architecture to detect genuine and forgery signatures by using particle swarm optimization | |
| dc.type | Resource Types::text::report | |
| dspace.entity.type | Publication | |
| oairecerif.author.affiliation | Universiti Sains Malaysia |