Abstract:

The authors proposed a method for constructing neural network ensemble based on affinity propagation and genetic algorithm, which can be  applied to handwritten digit recognition. Firstly, we extracted features of samples from the MNIST handwritten digit database by PCA (Principle Component Analysis) and Fisher LDA (Linear Discriminant Analysis) respectively. Secondly, we adopted the bagging method to train some BP neural networks as candidate networks using the extracted PCA and LDA feature sets. Thirdly, affinity propagation algorithm was used to group candidate networks and find exemplars in each cluster. Finally, we optimized these weights for exemplars using genetic algorithm and integrate exemplars with the optimized weights to obtain the final classifier. Experimental results on the MNIST handwritten digit database prove that the accuracy of the algorithm proposed in this paper is far higher than that of individual network while its efficiency is acceptable.

Key words: artificial neural network, handwritten digit recognition, neural network ensemble