Image Classification Using Optimized Synergetic Neural Network

Abstract

In this paper an empirical study on image classification using optimized synergetic neural network is conducted. Unbalanced attention parameter type of synergetic neural network is enhanced by applying optimization algorithms such as particle swarm algorithm, quantum particle swarm algorithm and Shuffled Complex Evolution with PCA. In this work Gabor-wavelet algorithm has been applied in the feature extraction stage. The ABM image dataset consisting of four classes of Bear, Cat, Cow and Wolf is used and divided into two groups of training and test sets. The aim of this empirical work is to optimize unbalanced attention parameters of synergetic neural network, to achieve the highest accuracy of image classification. Results are calculated after applying three-fold cross validation. According to the results, optimized synergetic neural network performed better than the simple synergetic neural network. The highest classification results obtained using SNN-QPSO, which is 0.94. © Springer-Verlag Berlin Heidelberg 2013.