Farsi handwritten digit recognition based on mixture of RBF experts

In this paper, a new classifier combination model is presented for Farsi handwritten digit recognition. The model is consisted of four RBF neural networks as the experts and another RBF network as the gating network which learns to split the input space between the experts. Considering the input data, which is an 81-element vector extracted using the loci characterization method, the gating network assigns a competence coefficient to each expert. The final output is computed as the weighted sum of the outputs of the experts. The recognition rate of the proposed model is 93.5% which is 3.75% more than the rate of the mixture of MLPs experts previously ran on the same database.