An Effective Learning Algorithm of Synergetic Neural Network

In Synergetic Neural Network (SNN), the learning problem can be reduced to how to get prototype pattern vector and adjoint vector. Here we put emphasis on the study of the former. A novel-learning algorithm of SNN is presented in this paper, which combines the self-learning ability of SNN with the global searching performance of Immunity Clonal Strategy. In comparison with learning algorithm on superposition of information (LAIS), the proposed method can avoid local extremum and improve searching efficiency. Experiments show that the new method not only overcomes the shortages of methods available but also enhances the classification accuracy rate greatly.