Evolutionary structure optimization of hierarchical neural network for image recognition

The purpose of this paper is to optimize the structure of hierarchical neural network (NN). In our proposed method, the structure optimization is considered as combinatorial optimization problem, and unnecessary connections in trained NN are eliminated by using genetic algorithm (GA).We focus on the NN which specialized for image recognition problems. In order to validate the usefulness of the proposed method, face recognition and texture classification examples are used. From the experimental results, it was shown that compact neural network was generated, keeping generalization performance by proposed method.