X - International Journal of Information Science and Computer Mathematics (Closed Ed TRF)
Volume 2, Issue 2, Pages 129 - 136
(November 2010)
|
|
FARSI HANDWRITING RECOGNITION WITH MIXTURE OF RBF EXPERTS BASED ON PARTICLE SWARM OPTIMIZATION
Mohammad Javad Abdi and Hamid Salimi
|
Abstract:
In this paper, a classifiers combination based on Particle Swarm Optimization (PSO) is presented for Farsi handwritten digit recognition. This model consists of four Radial Basis Function (RBF) neural networks as experts and another as the gating network which tries to separate the input space among the experts. PSO is used to train RBF neural networks (NNs). Input data is an 81-element vector. This vector is extracted using loci characterization method. It is shown that recognition rate of the proposed method is 97.1% which is 1.8% more than the rate of mixture of RBF experts previously ran on the same dataset. |
| Keywords and phrases: computational efficiency, handwritten digit recognition, loci characterization method, mixture of experts, particle swarm optimization method, radial basis function. |
| Communicated by Kewen Zhao |
| Number of Downloads: 54 | Number of Views: 209 |
|
